University of Southern California Dissertations and Theses

Development of an integrated biomechanics informatics system (IBIS) with knowledge discovery and decision support tools based on imaging informatics methodology

(USC Thesis Other)

Development of an integrated biomechanics informatics system (IBIS) with knowledge discovery and decision support tools based on imaging informatics methodology

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Development of an integrated biomechanics informatics system (IBIS) with
knowledge discovery and decision support tools based on imaging informatics
methodology

By

Joseph William Liu

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
BIOMEDICAL ENGINEERING

August 2022

Copyright 2022 Joseph William Liu

ii

Acknowledgements
Thank you for the guidance, collaboration, and contribution of the following advisors and
colleagues:
Brent Liu, Ph.D. Associate Professor, Advisor
Jill McNitt-Gray, Ph.D. Professor, Advisor
Ximing Wang, Ph.D. Mentor, USC IPI Lab
Sneha Verma, Ph.D. Mentor, USC IPI Lab
Kevin Ma, Ph.D. Mentor, USC IPI Lab
Casey Wiens, Ph.D. Candidate, USC Biomechanics Lab
Harper Stewart, Ph.D. Candidate, USC Biomechanics Lab

iii

Table of Contents

Acknowledgements……………………………………………………………………………..... ii
List of Tables……………………………………………………………..…………...………...... v
List of Figures………………………………………………………………………...………...... vi
Abstract..……………………………………………………………………...……………........ vii
Chapter 1: Introduction………………………………………………………………….……..…. 1
1.1: Human Performance Research………………………………………………………. 1
1.2: Data Management in Biomechanics Research in Lab/Field/Large Scale Trials…….. 4
1.3: Limitations and Need For An Informatics System…………………………………... 6

Chapter 2: Background…………………………………………………………………………… 9
2.1: Biomechanics Research Workflow………………………………………………….. 9
2.2: Medical Imaging Informatics Based Principles Applied to Human Performance
Research and System Development………………………………………………... 14
2.3: Previous Work: Imaging Informatics-Based Workflow Engine and WADO
Viewer……………………………………………………………………………... 23
2.4: Previous Work: Solutions for Human Performance Analytics………….………….. 24

Chapter 3: Integrated Biomechanics Informatics System (IBIS) Based On Medical Imaging
Informatics Core Principles……………………………….………………………… 27
3.1: System Architecture and Workflow…………………….………………………….. 27
3.1.1: Data Collection………………………………….…...…………………… 28
3.1.2: Data Preprocessing with Support Tools…...…….…………..……………. 30
3.1.3: Data Model/Organization……………………….………………………... 34
3.1.4: Data Upload……………………………………………….……………… 35
3.1.4.1: Web Server/Web-Based User Interface…………………….…... 36
3.1.4.2: Auto Uploader/Parser…………......……………………………. 39
3.1.4.3: Remote Access…………………………………...…….………. 40
3.1.4.4: Security/User Authentication………………………………...… 42
3.1.4.5: Privacy/User Access……………………………………………. 42
3.1.5: Storage/Management…………………….……………………………….. 44
3.1.5.1: System Admin Control Panel..……………………….…………. 44
3.1.5.2: Data Anonymization……………………………………………. 45
3.1.5.3: User Access Controls/Permissions……………………………... 46
3.1.5.4: Data/Metadata Storage…………………………………………. 47
3.2: Knowledge Discovery and Data Analysis Tools…………………………….……… 48
3.2.1: Data-Viewer Application………………………………………….……… 49
3.2.1.1: Synchronized Multimedia Data Analysis………………………. 50
3.2.1.2: Customizable Layouts………………………………………….. 52
3.2.2: Data-Mining and Decision Support………………………………………. 53
3.2.2.1: Integrated Statistical Analysis Tools……………………………. 54
3.2.3: Annotation/Communication……………………………………….……... 58

iv

Chapter 4: Use Cases in Biomechanics Research…………………………..…………………… 61
4.1: USATF Triple Jump and Long Jump Project………………………….……………. 61
4.2: PAC12 Long Distance Running Stress Fracture Study……………………………... 65

Chapter 5: Results/Evaluation…………………………………………………………………... 70
5.1: USATF Triple Jump and Long Jump Project………………………………………. 70
5.1.1: Data Collection/Preprocessing……………………………...….………… 71
5.1.2: Data Upload/Storage……………………………………….…...………... 71
5.1.3: Data Review/Analysis……………………………………………………. 72
5.1.4: Knowledge Discovery/Decision Support………………….……………… 77
5.2: PAC12 Long Distance Running Stress Fracture Study…………….………………. 80
5.2.1: Data Collection/Preprocessing…………………………………………… 81
5.2.2: Data Upload/Storage……………………………………………………... 82
5.2.3: Data Review/Analysis……………………………………………………. 83
5.2.4: Knowledge Discovery/Decision Support…………………….…………… 85
5.3: Workflow Efficiency Comparison………………………………………………….. 89

Chapter 6: Discussion and Future Work……….………………………...……………………… 92
References………...…………………………………………………………………………...... 98
Appendix..……..…...………………………………………………………………………...... 100

v

List of Tables

Table 4.1: USATF Study Data Types and Formats…….………….…………...……………..…. 62
Table 4.4: PAC12 Study Data Types and Formats…………………..…………………………... 67
Table 5.1: Workflow Tasks……………………………………….……………………………... 90

vi

List of Figures

Figure 2.1: Biomechanics Research Workflow…………………………………………………. 10
Figure 2.2: Medical Imaging Informatics Five Layers……………………………………..….... 15
Figure 3.1: IBIS Workflow………………………………………...……………………………. 28
Figure 3.2: Automatic Foot Detection Algorithm……………………………………………….. 32
Figure 3.3: Data Model…………………………………………………………….…………..... 35
Figure 3.4: Web-Based Graphical User Interface……………………………………………...... 37
Figure 3.5: Subject Upload Interface………………….……………………...………………..... 38
Figure 3.6: General Uploader……………………………………………………...…………..... 39
Figure 3.7: Gateway/Parser……………………………………………….……………….…..... 40
Figure 3.8: Centralization of Data………………………………………………………………. 41
Figure 3.9: User Access Controls……………………….…….……………..….….….………... 43
Figure 3.10: System Administrator Control Panel………………………………………..……... 46
Figure 3.11: Built-In Data Viewers……………………………………………………………... 50
Figure 3.12: Synchronized Multimedia Viewing Dashboard………………….………………... 52
Figure 3.13: Customizable Layout Template…………………………………….……………... 53
Figure 3.14: R Shiny Integrated Statistical Analysis Tool……………………………….……... 55
Figure 3.15: Cross-Study Analysis…………………………………………….………………... 57
Figure 3.16: Annotation and Collaboration Module……………………………………..……... 59
Figure 4.2: USATF Research Study Workflow………………………………………..……….. 63
Figure 4.3: PAC12 Research Study Workflow………………………………………..………... 66
Figure 5.1: USATF Workflow Comparison…………………………………………………….. 70
Figure 5.2: USATF User Interface……………………………………………….……………... 73
Figure 5.3: USATF Simultaneous Data Analysis…………...…………………………………... 75
Figure 5.4: USATF Data Viewer Layouts………………….…………………………….……... 76
Figure 5.5: USATF Longitudinal Data Analysis….……………………………………………... 80
Figure 5.6: PAC12 Workflow Comparison……………………………………………………... 81
Figure 5.7: PAC12 Data Upload/Sharing………………………………………………………... 84
Figure 5.8: PAC12 R Shiny Data Analysis……………………………………………………... 86
Figure 5.9: PAC12 Cross-Study Data Analysis…………………………………..……………... 88

vii

Abstract

The field of biomechanics involves integrating a variety of data types such as waveform, video,
discrete, and performance. These different sources of data must be efficiently and accurately
associated to provide meaningful feedback to athletes, coaches, and healthcare professionals to
prevent injury and improve rehabilitation/performance. There are many challenges in
biomechanics research such as data storage, standardization, review, sharing, and accessibility.
Data is stored in different formats, structures, and locations such as physical hard drives or
Dropbox/Google Drive, leading to issues during sharing and collaboration. Data is reviewed and
analyzed through different software applications that need to be downloaded and installed locally
before learning how to use. An integrated biomechanics informatics system (IBIS) built based on
the core principles in medical imaging informatics provides a solution to these many challenges.
The system provides a secure web-based platform that will be accessible remotely for
authenticated users to upload, share, and download data. The web-based application includes built-
in data viewers that are streamlined for reviewing multimedia data and decision
support/knowledge discovery tools. These tools include automatic foot contact detection for pre-
processing, built-in statistical analysis applications for longitudinal and cross-study analysis, and
a multi-institutional collaboration module. The IBIS system creates a centralized hub to support
multi-institutional collaborative biomechanics research and analysis that is remotely accessible to
all users including athletes, coaches, researchers, and clinicians generating a novel streamlined
research workflow, data analysis, and knowledge discovery process.

1

Chapter 1: Introduction
In this proposal, I present the development of a multi-institutional collaborative platform
for sports performance and injury prevention research that is based on the principles of imaging
informatics. The sports data from biomechanics labs and sports research groups may come in a
large variety of forms and they can also be organized in different ways. The proposal will go over
the approaches used in developing a web-based informatics system that allows for secure upload,
download, and management of data. I will also discuss the built-in data viewer and knowledge
driven statistical data analysis tools and the evaluation in use cases.
In order to understand injury prevention and performance improvement in sports, it is
necessary to have a strong understanding on the biomechanics of the human body. This includes
analyzing the size/dimensions of the different body parts, the joint angles at which each of the
components are connected to each other, and how these different parts work together to create a
movement and exert a force. This additional layer of complexity involves recording and analyzing
the speeds and directions the many body components are moving in, which allows us to study
forces, torque, momentum, and more. The study of these forces allows us to be understand how
different areas of the body can work together to generate compound movements, make a movement
more efficient, or interact with its environment. In depth analysis requires the background of
anatomical and biomechanical knowledge combined with a method of recording the movements
and associated measurements for review.
1.1 - Human Performance Research
Research in the field of sports performance and injury prevention includes the use of a
variety of multimedia data. This data can be made up of images, video, force data, and more that

2

are then studied and analyzed to look at athlete movements in detail and then provide feedback to
improve technique for performance or for safety
5
. The collected data are used along with
knowledge in biomechanics to study and analyze the movement of the human body. The
biomechanics of the body and the limbs, which includes the timing, the direction, and the speed of
the movements, are recorded through video. Along with the video, force plates can be placed on
the ground that can measure the amount of force exerted by the athlete either during takeoff on a
long jump or a stride during a run. By combining the video and this force data, a more
comprehensive view can be obtained and studied
4,5
. The combined recorded data can be analyzed
together with the knowledge of biomechanics to get a deeper understanding of what is happening
at specific point in time during an athlete’s movement. The next step is to then use this new
knowledge along with the coaches and athlete expertise/experience in their specific sport to
achieve certain results such as preventing injury and improving performance.
The majority of the data that is collected in human performance related research in sports
is of the individual athlete. The examples will be used cases in both track and field long jump,
triple jump, and also long-distance running. The data is collected on stand-alone devices which
may be different at each collection location and not integrated with each other. Once this data for
an individual athlete is collected, it can also be compared with the data from another individual
athlete. In this way athletes’ data can be compared across teams/institutions/research groups. For
example, for a specific event such as the long jump, trials from the same athlete can be compared
side by side to study the events that led to better performances or events that caused pain and could
be avoided to prevent injury. This can be used to monitor any changes in the athlete’s performance
or technique longitudinally across time by comparing with previous collected data. Once data is
collected for different athletes, their individual data can then be put together and compared in

3

studies to discover if there any correlations or common factors that affect their safety and/or
performance.
The applications of human performance research and sports data have great potential in the
future. Another step further would be to collect video and other kinds of data on multiple athletes
concurrently. One example is to record and analyze the movements of many athletes or a team as
a whole. Video data could be analyzed to study athlete the movements of the athletes and see if
particular placements of players led to a better performance as a team.
Injury Prevention
By studying the biomechanics of an athlete’s body and movement researchers and
scientists are able to make discoveries that give insight into how injury or reoccurrence of injury
can be prevented. These discoveries could be something such as a key event where a body part or
collection of body parts moves in a certain way and in turn causes the athlete pain or discomfort.
The first step is to discover what the event or events are undesired movement that may lead to pain
or injury. Then once that is identified, the collected biomechanics data such as force data and video
provide more detailed information about the movements of the body during those events, including
the athlete’s trajectory, speed, and more. Studying these details allows us to accomplish two tasks.
The first thing is to study these details to better understand why and how these movements in the
identified event occurred. This understanding leads to the second task which is identifying a related
movement that can move the body in a similar desired fashion but without causing pain or injury.
This can be extended to multiple areas in rehabilitation, such as in wheelchair subjects studying
the mechanics and rotations of the shoulder movement.
Performance Improvement

4

In addition to clinical applications, the analytics of sports related data can be used to
improve the performance of the athlete. Without the data collected from biomechanics research,
an athlete’s improvement in performance would be dependent on repeating multiple trials with
their coaches and reviewing their results together. By combining the coach’s expertise with
biomechanical data such as video and force data gives the coach and athlete in depth look and
extensive knowledge about the recorded movements. This raw data is then processed further by
the biomechanics lab creating processed data such as force vector overlay videos and strobe motion
images for even more sophisticated analysis of each of the athlete’s movements during each trial
and provides huge insight into how changes in the technique and preparation can lead to better
performance. Both the raw and processed data provided by biomechanics research groups are
essential tools that coaches and athletes can utilize together along with their experience and
expertise in the field. This combined effort of researchers, athletes, and coaches enhances the
ability to discover key events during the athlete’s movement that could be adjusted to improve
their performance.
1.2 – Data Management in Biomechanics Research in Lab/Field/Large Scale Trials
Biomechanics research is conducted by many research groups working in conjunction with
sports teams and organizations. Depending on the collaborating institutions, collections can differ
in scale and location, for example, in the research lab, in the field, or training site. These varying
factors affect the research workflow and the tools and resources available. Sports data is collected
at different field sites in different types and formats
4
. Because the data is collected on varying
standalone systems, not only are different formats collected, data is collected in different
combinations based on research practices or preferences and also processed uniquely. Add also
that data is collected standalone and store standalone and processed standalone.

5

It is of critical importance to store sports performance and biomechanics data collected
from different athletes in a secure and efficient manner
1
. There are many different methods, each
with its own advantages/disadvantages, of storing data currently used including Google Drive,
Dropbox, or physical hard drives
4
. Data sharing is also an essential element in biomechanics
research to acquire larger sample sizes and allow collaboration between multiple groups of
athletes, coaches, other researchers. To share data using different applications such as Google
Drive and Dropbox, access rights for files/folders and user privileges have to be set in each
application that users need to download and use. Data organization is another issue that arises
during collaboration. One research group may organize their collected data primarily by athlete
while another group may store everything primarily by collection. These methods are tailored to
the needs and workflow of each research lab and devices available at collection sites.
Data collected at the collection site is considered raw data that has not been altered or
processed through any code or algorithms
4,5
. Raw data includes video, force, acceleration data
acquired at the collection saved locally on a computer or hard drive. Once the data raw data is
collected and backed up, there are steps where the data is preprocessed and cleaned up in
preparation for the complex processing in the following steps
4,5
. These steps are unique to each
group and specific collection sites. Before this data can be shared between groups, the data needs
to be standardized. There may local measurements and adjustments that need to be made according
to the specific factors of each collection environment.
There are many important preprocessing steps including cleaning data such as cropping
videos down to smaller sizes to only include the portions that will be analyzed as longer videos
and larger file sizes lead to storage issues and slow code processing time. Another essential
preprocessing step is identifying key events, such as the athlete’s foot’s first contact with the force

6

plate to sync video and force data. This time sync leads to further processing steps and more
advanced processed data such as vector overlay images and videos for deeper analysis. There is a
complex workflow of data from its raw native form to preprocessed data and then finally
postprocessed data. When the data is ready for review/analysis, the data files such as videos and
images are opened with different software. Communication between the biomechanics researcher
and the coach/athlete is done in-person at meetings during collections/events or online through
email.
1.3 - Limitations and Need For An Informatics System
The current workflow in the space of sports performance and biomechanics research is
complex because every essential element from data collection to data processing and review is
very specialized. Each step in the workflow comes with its own challenges of accessibility,
efficiency, and security.
The first challenge is during the data collection. If there are different types of data and then
different formats for some of these types, this will pose a challenge when trying to view the data
later. For example, to view collected raw videos that are of different file formats, there is a need
to open different programs on the computer and wait for them to load while managing the window
sizes of each to view them simultaneously. And if the data is located on two standalone systems,
the data would need to be opened and viewed on two separate systems. An integrated informatics
system with a universal video viewer built in will be able to fix these issues and also allow for
streamlined simultaneous review of multiple videos at the same time. The use of an informatics
system and a gateway/parser will also aid in the standardization of numerical data that will be

7

studied and shared among collaborating groups, which leads to the next topic of data upload and
storage.
The second difficulty arises during the storage and management of data during
collaboration. Different storage methods for data come with their own advantages but also some
shortcomings. A hard drive that is encrypted would be secure but because it is located on the
physical device it is hard to share with collaborators who are located far away. An integrated
informatics system will contain a secure database that users would need to be authenticated to log
in to and upload data. The database will be a centralized location where all the data will be stored
with a standardized organization. There would be no need for different research groups using
different programs such as Google Drive and Dropbox having to each download both programs to
share data. Also assigning specific access rights and permissions for the data would have to be
done in each of these separate programs. The centralized database in the informatics system will
have a web-based user interface that would allow authorized users to access the data remotely
through the internet and download or review the data. The advantages of the informatics system
would allow for remote access capabilities which allows for efficient data sharing/collaboration
and also give researchers/athletes/coaches flexibility in when and where they would like to upload
or download whether it is at the collection site, sports meet, hotel, on the go, or in the lab. All the
data and multimedia files will be available for viewing without the need for any extra software to
be downloaded and installed.
The third area in which an integrated informatics system can improve current workflows
by providing a platform for rapid knowledge discovery of data collected as well as future data
processing and analysis tools. Current preprocessing and processing steps involve human decision
making. For example, to run a specific code analyzing an athlete’s jump, the time the athlete’s foot

8

first contacts the ground is an essential key event and a researcher may go through the slides to
identify the frame of the video in which it occurs. If this is done for each trial, with multiple trials
per collection, and multiple collections for each athlete, this leads to a large amount of time
allocated for this key step. One of the support tools that will be built for the informatics system
can help either automate the process or automatically identify a set of frames where the foot contact
could occur, allowing the researcher to only have to choose from a few slides which would
drastically reduce the amount of total time spent. Other support tools include statistical analysis
tools that will be built into the informatics system to fix the issue of needing to open and run a
separate statistical analysis program. With analysis tools built into the system, all the data and tools
needed for review and analysis will be centralized in one location. By centralizing the data in one
location and providing a method of standardization/normalization, data analysis and knowledge
discovery can be achieved much quicker and with high efficiency.
In my proposal, I will develop an integrated biomechanics informatics system, also known
as IBIS for the biomechanics and sports research workflows that will provide a platform for
centralized data storage/management, standardized organization, secure remote accessibility, and
integrated advanced support tools for deep analysis and knowledge discovery. With this system,
all users in the biomechanics workflow, including researchers, clinicians, coaches, and athletes,
will have a streamlined platform for collaboration that combines the various expertise to analyze
the data with novel methods and discover the right questions to ask and key areas to explore.
In order to develop this system, it is essential to draw from core principles in the medical
imaging informatics workflow, including standardization, security/privacy, accessibility, which
will be discussed in the next chapter.

9

Chapter 2: Background
2.1 - Biomechanics Research Workflow
By studying both the biomechanics research and medical imaging informatics workflows
I was able to draw parallels and identify core principles from medical imaging informatics
systems that can be applied to the biomechanics research space. The biomechanics research
workflow consists of key essential steps described and shown in Figure 2.1 below. I will go over
each step in detail in the following section.
1. Data Collection (Figure 2.1 - Number 1)
Currently, data collected at different field sites for sports and biomechanics research are
acquired in many different types and formats
4
. These differences stem from the types of data that
can be recorded including video data, force data, acceleration data, and factor such as video view
and zoom. Different groups can collect different combinations of these videos based on their own
research practices or preferences. The collected video data then may also be in different formats
such as .flv, .mp4, etc depending on the recording devices used and those available at the different
collection sites. As an example, one research group that is collaborating with another may choose
to collect and analyze GPS data attached to a collection, while another group may choose to focus
on the other types of data. Once this data is pre-processed and then post-processed, the results will
also vary in format because each institution may have their own code for processing and analysis.
Challenges may also occur in the definition of key elements in the data workflow. For
example, in the same sport, there could be differences in the definition of trial. In a study for long
distance running and the causes of stress fractures, one institution has a set-up where each time the

10

athlete steps on the force plate is defined as a trial, whereas another collaborating institution has a
treadmill setup where a series of multiple steps is defined as a trial.

Figure 2.1: The main key steps in the overall biomechanics research workflow: 1) data collection,
storage, 2) data cleanup and pre-processing, data post-processing, 3) storage and 4) data
review/analysis and sharing/collaboration. The complexity and non-linearity of the workflow is a
result of many unique factors in biomechanics such as the nature of the sport, the collection site,
available devices and resources, time available, etc.

2. Data Preprocessing/Postprocessing (Figure 2.1 - Number 2)
Data that is collected by biomechanics research groups at the collection site is considered
raw data, data that has not been altered or processed through any code or algorithms
4,5
. Raw data
would include video, force, acceleration data acquired at the collection and saved locally either on
a computer or a hard drive. Once the data raw data is collected and backed up, there are steps
where the data is preprocessed and cleaned up in preparation for the complex processing in the
following steps
4,5
.
1
2
3
4
2

11

Some preprocessing steps are unique to the biomechanics lab and their specific collection sites.
Depending on the hardware used, video data may be collected at different frame rates at each
collection site by different collaborating research groups. Before this data can be shared between
groups, the data needs to be standardized beforehand. There may also be local measurements and
adjustments that need to be made according to the specific factors of each collection environment.
For example, in the analysis of steps and foot strikes on a force plate in the ground or in an
instrumented treadmill, there are inherent differences and adjustments that need to be accounted
for
14,15,17
. For movement on instrumented treadmills there will be associated artifacts in
measurements as well as differences in velocities of the center of gravity compared to overground
running
18,19,20
.These preprocessing steps to clean or standardize the data are based on lab setups
and methods are essential for further processing and group collaboration between multiple research
institutions.
Other important preprocessing steps include cropping video files down to smaller sizes. To
ensure everything is collected during a trial and nothing is missed, the cameras start recording
before the athlete begin to move and end recording a bit after the athlete has finished their
movement. There is extra time in the video before and after the athlete’s movement where there is
nothing important occurring. This creates longer videos and larger file sizes leading to storage
issues and slowing processing steps by adding more time the codes/algorithms have to run.
Identifying key events in the data, such as the athlete’s foot’s first contact with the force
plate, is key for identification used in subsequent processing steps. A preprocessing step is to run
a code for a user to identify frames during a video where foot contact occurs and outputs a frame
number and time stamp used to sync video and force data. The time sync mentioned above leads
to further processing steps and more advanced processed data such as vector overlay images and

12

videos for deeper analysis. Measured forces in multiple dimensions are used to produce force
vectors and their angles
16
. Vector overlay videos show an athlete’s movement with the force vector
measured by the force plate underneath overlaid on the athlete in real time. This allows the
researcher, the coach, and the athlete to simultaneously watch the video recorded of the specific
trial and view real-time measurements and direction of associated force data. The delicate and
detailed workflow of data from its raw native form to preprocessed data and then finally
postprocessed data is what enables the research groups to accurately and efficiently create
insightful data for next level analysis.
3. Data Storage/Management (Figure 2.1 – Number 3)
Data storage and management is a unique area within the biomechanics workflow in that
it can be performed at multiple stages during the workflow. Depending on various factors such as
the location of the collection site, the time available, or resources available, data may need to be
stored right after its collected for pre-processing at a later time. Raw data can be cleaned and sent
to the coaches and athletes for review immediately while further processing and analysis is done.
Sports performance and biomechanics data are collected from different athletes from multiple
collection sites and so it is critical to store any collected raw or processed data in a secure and
efficient manner
1
. There are many different methods of storing data that are currently used. Some
research groups use Google Drive, some use Dropbox, while others keep all the data on hard
drives
4
. Certain methods have their own advantages/disadvantages. For example, the use of a hard
drive is secure because one would need to physically have the hard drive to access or copy the data
and also the hard drive could be passcode encrypted. However, sharing the data across many
collaborating research groups located in different locations by physical transport of the hard drive
would be an issue. Using third-party resources such as Google Drive or Dropbox have its

13

advantages but also its own issues. Users would need to learn how to use the platforms to share
files, assign access rights, etc. Collaborating groups may use different ones and would need to
download and/or learn to use these different platforms.
Another issue lies in that collaborating research groups may not organize files and folders
in the same structure. One research group may organize their collected data primarily by athlete,
then inside the main athlete folder, there are multiple folders for each of the athlete’s collections.
In these folders for each collection there are individual folders for each trial which includes all of
the collected data for that specific trial. Another group may store everything primarily by collection
folders. Each folder for a collection would have athlete folders for each athlete that participated in
that collection, and then inside these athlete folders there would be individual trial folders with the
associated data for that trial. There is no exact way to determine whether one method is more
“correct” than the other because these methods are tailored to the needs and workflow of each of
the sports research labs. The devices available at each collection site and the time constraints
during collection are all essential factors affecting methods of data storage used.
4. Data Sharing/Collaboration (Figure 2.1 – Number 4)
Data sharing and collaboration is a very important element in biomechanics and sports
performance research. In order to acquire larger sample size, data is collected from different
collection site/locations for more accurate analysis. After the data is collected at a field site it is
stored locally on a computer/laptop and saved onto an encrypted hard drive. Data can be shared
from the hard drive with collaborating research groups at the collection site or after the research
group has returned to their lab. The security of using the hard drive is that the physical drive must
be present in order to access the data to review it or make copies for a collaborator.

14

Another method for data sharing and collaboration is through online third-party
applications such as Google Drive/Dropbox. These applications allow for data to be uploaded into
folders and subfolders and then shared with athletes, coaches, other researchers who are in other
locations with internet access. Specific access rights for each file/folder have to be set in each
application for the specific user email. Detailed user privileges such as reading, editing, deleting
data are individually customized as well. The advantages of using this web-based method are that
data is not restricted to a physical location like a hard drive. However, different applications aren’t
linked together. To view a folder of data located in Google Drive, a user with shared access must
reach the data through Google Drive. To access a shared file on Dropbox, the user would need to
download and install the Dropbox application. There are many methods used for collaboration in
the biomechanics and sports data field, each one with their advantages but key differences.
2.2 - Medical Imaging Informatics Based Principles Applied to Human
Performance Research and System Development
The overall goal of this proposal is to create a system that uses the key core principles in
medical imaging informatics that improves the current human performance and biomechanics
research workflows. Although the biomechanics and medical imaging research spaces are different
disciplines there are parallels between essential workflow steps including data storage,
management, review/analysis. By studying these similarities, I found many areas where a medical
imaging informatics-based system can provide many advantages such as efficiency, security,
privacy, accessibility, and built -in analysis tools. These advantages are derived from the five-layer
Medical Imaging Informatics Infrastructure (MIII) diagram shown below in Figure 2.2, which will
help in studying the parallels between medical imaging informatics and biomechanics research.

15

Figure 2.2: The 5 layers (in order from bottom to top) of the Medical Imaging Informatics
Infrastructure (MIII) encapsulate the different components involved in the development and
framework of clinical and research applications.
6
The first parallel between biomechanics research and medical imaging starts with the data
acquisition stage and data storage, which is the bottom layer of the MIII in the Figure 2.2. Similar
to how medical images may be collected from different devices at different remote locations,
biomechanics data comes from different devices and collection sites as well. Each collection site,
whether it is a training facility or a trial competition location, may have different hardware set up,
which collects data not only in different formats but also stores the data in varying organizational
structures. Inspired by the DICOM movement that standardized medical images across different
institutions from all over the world, the goal in biomechanics is to aim for standardization as well
to improve efficiency when storing, managing, and sharing data. In collecting biomechanics data,
the data formats and types may differ due to the available devices and resources available at each
collection site, however, important standardization can be accomplished in the data organization
stage. By having collaborating research groups adopt a similar data model and subfolder
organizational structure, uploading into a centralized location together with any collaboration will
be improved.
This leads to our next parallel of accessibility. Collaboration in the medical imaging
clinical workflow is needed as clinicians review images along with the readings from radiologists

16

and other associated groups. Similarly, in the biomechanics research workflow data is shared and
analyzed between collaborating research groups, coaches, and athletes. In medical imaging
informatics the PACS server and database acts as one centralized location where medical images
and patient information are stored. This data can be visualized and analyzed on various
workstations and remotely from web-based applications making it easier for all collaborating
clinicians, making up the second layer (from the bottom) of the MIII in Figure 2.2. The
biomechanics informatics system will include a database and acts as a centralization location
where data from different locations will be uploaded and stored. Then from a web-browser based
application the data can be accessed remotely allowing much greater accessibility than previous
solutions of sharing physical hard drives and downloading/installing different third-party software.
Another part of the MIII second layer (Figure 2.2) common to both medical imaging
informatics and biomechanics research is the need for security and privacy of the collected data.
In the clinical setting a patient’s medical history and medical images are of utmost important and
ensuring that the data is secure and only accessible by authorized personnel is essential. It is
necessary to treat an athlete’s sports data with the same security and privacy, ensuring that only
the data is only accessible to the athlete themselves, their coaches, and any collaborating
researchers. In the medical imaging informatics workflow this security/privacy is protected within
the PACS server and database, which have user authentication protocols, access privilege controls,
and firewalls put in place. In a similar manner a biomechanics informatics system where sports
data is stored, managed, and accessed will also include similar user authentication protocols,
specific user access controls, and protective firewalls.
The web-based applications in imaging informatics workflows include built-in knowledge
discovery and data mining/analysis tools to help users review the images/data and then provide

17

some sort of analysis which they can store as annotations back into the database (MIII third layer
in Figure 2.2). There are also various computer-aided detection (CAD) algorithms run on images
to provide calculated results and further insight on the medical images. Drawing inspiration from
these clinical tools, built-in tools were developed to aid in the biomechanics research workflow.
These tools include an annotation/collaboration module that can be used to for collaboration during
data analysis, data pre-processing tools, and statistical analysis tools.
The MIII fourth layer in Figure 2.2 encompasses three area of focus for the user application.
The first area of research in medical informatics includes clinical trials and knowledge discovery,
which is similar to the research application in biomechanics for longitudinal studies and
discovering correlations with injury. The second area of clinical service in medical imaging
informatics to help improve clinical workflows and efficiency is parallel with the improvement of
data visualization and review workflows for the coach and athlete users. The third area of education
and training in the clinical workflow is similar to performance enhancement applications for the
athlete and coaches in biomechanics. This layer requires knowledge from experts on each of the
subject matters to develop a system tailored for specific user needs. The fifth layer of the MIII
shown in Figure 2.2, is the integration of the entire system and user applications for medical
imaging informatics. Likewise, a biomechanics informatics system provides an overall
architecture that streamlines the integration of user applications.
In carefully studying the medical imaging informatics clinical workflow and infrastructure,
I have adopted its infrastructure and many of its key principles in building an informatics system
for biomechanics research workflows. Due to the nature of differences in the sports and
biomechanics research, the informatics system will need to be adaptable to fit its specific needs,
however, there are many parallel aspects between the two workflows allowing us to build on the

18

core principles that already exist. By following the blueprint for the gold standard in the medical
field, I will develop a system with all the key components and push the boundaries further to fit
the needs of the field of sports biomechanics research.
Electronic Patient Record (ePR) Framework for Biomechanics
The electronic patient record (ePR) is an advanced digital folder system of clinical
information and different forms of data that is patient-based
9
. The ePR includes patient health
information as well as imaging data and is not just limited to an information model
9
. It also includes
web-based user interfaces, security measures, and decision support tools that are built into the
application
9
. The patient-based design allows for users to select the specific patient and all of their
health data, medical images, and other related data such as radiology readings will all be in one
centralized location. The patient can be selected on an easily-accessible web –based application
that is secure and required user authentication. Once the patient has been selected, then the user
can navigate to the subgroup of data they would like to view including medical images, blood test
results, etc. With this design users in the medical imaging informatics workflow can easily locate
the patient and the exact information/data needed.
In studying various biomechanics research workflows, I identified an important parallel
between medical imaging informatics and biomechanics research: the advantage of having a
patient or athlete-based system. In biomechanics research there are key steps during the workflow:
data collection, storage, pre-processing, post-processing, visualization. In the current collection
workflow, the organization of data during the collection is dependent on the devices, hardware,
and time available at the collection site. Once the data collected and taken back to the lab for
storage and pre-processing, files and folders are organized with main folders for each athlete.
Inside these athlete folders there are folders for each collection, and in the collection folders there

19

are subfolders for different data types collected that contain the data files. This data organizational
structure is essential in biomechanics labs because data can then be easily located for post-
processing and then stored back into the system in an organized manner. This data model also
translates directly to visualization where an athlete’s raw and processed data can be easily located
to create/store reports and sent to coaches and athletes for review and analysis. The name adopted
for this athlete-centric system is the Electronic Athlete Record (eAR). One important goal was to
implement an athlete-based data model that could be utilized from the beginning of the
biomechanics research workflow and so an ePR and patient-based framework was built into the
design and development of the integrated biomechanics informatics system.
PACS (Picture Archiving and Communications Systems)
The development of PACS (Picture Archiving and Communication Systems) brought
about the development of new digital technologies used in the medical field such as the movement
from physical x-ray film to digitized x-ray images that could be transferred and stored on a
computer
6
. Along with the development of new digital technologies, PACS also jumpstarted the
development of electronic hospital/patient record systems used the clinical setting
6
. The field of
medical imaging informatics include everything from how medical images are acquired and
transferred to where and how they are stored along with other patient information. Advances in
medical imaging informatics allow for medical images and data to be transferred quickly within
hospital from department to department or from an off-site clinic to a hospital system. However,
with quick advances in medical technology, there were new issues of standardizing the medical
images and associate data, securing the privacy of the medical data to be transferred and stored,
and ensuring the safe access and review of the data. In order to address all of these issues and
ensure security, privacy and efficiency, informatics systems were built following key imaging

20

informatics core principles. The IBIS provides user authentication to protect the security of the
biomechanical data. User access controls are efficiently managed to ensure that privacy of the data
is maintained.
Standardization (DICOM)
One of the first key core principles in imaging informatics is standardization. This can refer
to the standardization of the formats of the data and how the data is organized before and after
being parsed into an informatics system. Taking one example from imaging informatics, after the
transition from X-ray films to digitized images, it was essential to create a single standardized
format for medical images so image data could be transferred between devices/computers and also
from clinics to hospitals, between hospitals, etc. The DICOM standard is essentially a set of rules
for everything related to the medical images, from the details of the device and setting the image
was acquired with to the compression of the actual image data
6
. Because DICOM is a universally
adopted standard, it allows images acquired by devices from different manufacturers to be readable
on all DICOM-compliant workstations. This is essential for DICOM images created by imaging
modalities from different companies so images can be read at any medical institution without
issues and can be transferred efficiently between departments/facilities. DICOM is incorporated
with other important standards and protocols including HL7 (Health Level 7), IHE (Integrating the
Healthcare Enterprise), TCP/IP communication protocols
6
. The IBIS promotes standardization of
biomechanical data and organization. Shared data between collaborating research groups are
stored under a universal athlete-based data model where data from each collection, regardless of
which site or institution it was collected, can be retrieved in streamlined fashion.
Security/Privacy

21

Once the data has been standardized and ready to be transferred, the next key core
principles are security and privacy. The integrity/safety of medical images are just as important as
any patient information. The HL7 (Health Level 7) standard is essential for healthcare database
information and assists in the connection and interface between PACS informatics systems and
the HIS (Hospital Information System)/RIS (Radiology Information System)
6
. These components
are important to ensure that while data is acquired/transferred/stored/reviewed the connections are
secure and only authorized personnel are allowed access.
The HIPAA (Health Insurance Portability and Accountability Act) mandates the need for
security and patient confidentiality, and this is enforced within the infrastructure of the PACS
medical imaging informatics system
6
. In order to access any of the patient data and/or medical
images (and associated medical reports/readings), the user must have an authenticated
account/password and to prevent any misuse of data, specific access rights can be controlled by
the system admin such as limiting a user and preventing them from accessing specific information
tables and sets of images. The security and privacy of medical data is of utmost importance in a
system designed to streamline data acquisition, transfer, storage, and access. The IBIS contains
security features to protect the privacy of the data similar to the ePR system. It is an eAR
(Electronic Athlete Record) that stores all of the data in an athlete-centric model. The system also
includes a web-based user interface and as well as built-in decision support tools.
Centralized Storage/Management
Another key core principle in the medical imaging informatics workflow is the
centralization of the data and efficient management of the data. In the PACS imaging informatics
workflow the images are collected by the imaging modality and sent to the PACS server archive,
which is also linked to the RIS (Radiology Information System)
6
. The PACS server is a centralized

22

location where medical images are stored and can be managed. The medical images and any patient
radiological data/medical history can be viewed from clinical workstations which are linked to the
PACS server. This centralized storage concept is efficient because it eliminates the need to retrieve
data from different locations or from different systems. In the IBIS, all of the biomechanical data
is uploaded into a single centralized location. Data from collaborating research groups and remote
sites are shared and managed through a web-based user interface. The user interface allows users
to upload, download, edit, and manage the data stored in the centralized database.
Accessibility
The next key core principle is accessibility of the medical images and data. Once the images
have been stored in the PACS server, they can be retrieved from directly linked clinical
workstations. In a system with web-based architecture included, client software for accessing the
data is in the form of a web-based application
6
. In a PACS workflow with web-based capabilities
the web-based application is platform independent as it can be run on any computer or laptop with
a compatible web browser and stable internet connection. PACS informatics systems have
exponentially increased the speed and efficiency of the medical imaging workflow. A medical
image that is now acquired digitally can be transferred almost immediately instead of waiting for
an x-ray film to develop, packaging it, and transporting it to another location. Efficiency is further
enhanced by PACS systems with web-based capabilities as authorized users such as radiologists
and clinicians can view the medical images remotely along with any patient data and provide
readings/feedback sent immediately back to the PACS server/HIS/RIS. The IBIS is a web-based
platform with a streamlined user dashboard. The remote capabilities allow users to access the
platform from any location to share and review the data from any device with internet access.
Data Review/Analysis

23

A key component of in the medical imaging informatics workflow is the data review and
analysis. As mentioned in the previous section, in the PACS system, radiologists and clinicians
can easily access medical images and patient medical history on clinical workstations. On
workstations, radiologists can perform readings of medical image and add any annotations/notes
for the referring physician to look at. The readings and the notes are then saved back into the PACS
server and the HIS/RIS. This is important because readings and notes are now saved in one
centralized location that can be accessed by different clinicians/departments on any clinical
workstation or through the web-based application for review/analysis. The IBIS is a web-based
platform that includes built-in data viewers from reviewing multimedia as well as a specialized
data-viewing dashboard for reviewing multiple forms of data simultaneously. There are also built-
in tools for streamlined statistical analysis as well as annotation and collaboration.
2.3 - Previous Work: Imaging Informatics-Based Workflow Engine and WADO
Viewer
In a previous work by Dr. Ximing Wang, an imaging informatics-based system was
developed that included a workflow engine that supported imaging-based clinical trials
7
. The
IWEIS (Intelligent Workflow Engine) system was able to take core principles of imaging
informatics principles such as security, privacy, and accessibility and build a system to support
imaging-based clinical trials
7
. For clinical trials, usually a custom-built system or a combination
of systems are set up in order to meet the demands of the workflow, however the IWEIS system
provides a streamlined solution
7
. The novel workflow engine allows the user to create different
modules where the user could upload medical images and then download later on. This capability
is especially useful in the clinical trial setting as it allows the user to build a customized workflow

24

and set up the modules in the needed framework. The system is user-friendly as there is minimal
programming in deploying the data management system.
The system also included a Web Access to DICOM Objects (WADO) viewer that allowed
users to view DICOM images on the web as long as they have access to internet
7
. The WADO
viewer was tested with MRI images of the brain for stroke patients
7
. These images were uploaded
into the database and the connected web server hosts the web-based interface. In order for users to
access the user interface, they must log-in with an authorized username and password. The web
access to medical images as mentioned before is a huge advantage to any workflow as it allows
remote access to the images for review and analysis. The WADO viewer can load medical images
along with any related metadata associated with the image and the study
7
. There are also web tools
on the viewer such as a free-drawing tool that can be used by clinicians to draw contours around
objects inside the image

and also digital brain templates that can be overlaid on top of the image
for reference
7
.
The IWEIS system and WADO viewer provided an extension of the imaging informatics
workflow to clinical image-based trials. The system reduced the time needed to set up a custom
workflow system for the image-based trials while maintaining security/privacy and improving
accessibility/efficiency
7
. The proposed system for biomechanics research would build upon this
and have the capability to match the needs of numerous varying biomechanics research workflows.
The integrated biomechanics system will be compatible with video and other types of multimedia
file types such as force and acceleration data. The system will also provide standardization for data
organization and storage, a streamlined data-viewing module, along with built-in advanced
decision support and statistical analysis tools.
2.4 - Previous Work: Solutions for Human Performance Analytics

25

In a previous work by Dr. Sneha Verma, the study investigated the workflow for human
performance analysis including the trials conducted and the variety of data collected
8
. As
mentioned in previous sections the data for human performance and sports athletic performance
include different kinds of multimedia such as videos at different frame rates, excel log sheets, force
data from industry devices, and much more. These data are gathered from stand-alone systems
from different collection sites including rehab facilities or other clinical settings and because of
this the data gathered from different facilities come in different folder structures
8
. Depending on
what devices and hardware are available at the collection sites, data will also vary in the format as
well
,
and due to these issues, the task of integrating all of this data for research and analysis is a
challenge
8
.
Past database solutions have been used for storing human performance data, however,
because the data is coming in different formats and also subfolder organizational structures, there
is a large amount of file/folder user manipulation/organization that must be done before storage
can be accomplished
8
. If the dataset is extremely large, the time it would take for users to manually
go through all the data and organize would be extensive and would not be efficient.
The solution proposed in the study was to develop a system capable of associating metadata
with the collected human performance data and by doing so creating a system to more efficiently
parse the data into a standardized organizational structure
8
. Once the data is parsed into
standardized structure, the process for uploading into a database for storage will be much more
streamlined. The system was created based off of the designs in the medical imaging informatics
workflow to adapt the same advantages for use in the human performance analysis workflow
8
. The
developed system was evaluated with data sets from a movement analysis study and was able to
streamline the human performance analysis workflow by providing a standardized organizational

26

structure for collected data with the use of metadata and also a storage system to efficiently store
both the data and metadata
8
.
The proposed integrated biomechanics system will also include a standardized
organizational structure for the collected data that will come from a variety of collection sites and
institutions. The standardization will be expanded across multiple collaborating research groups
from different institutions. Another novel layer of standardization will be introduced into the
biomechanics research workflow for integrating raw and calculated values from different devices
and sources. The unique database system and methods will make it possible to standardize and
combine collected discreet and also continuous data from multiple institutions with their own
collection devices, storage methods, research workflow. The IBIS system will allow users from
these different institutions to upload data into a centralized location with a common streamlined
organizational structure for review with a built-in multimedia data-viewing module and integrated
decision support/statistical analysis tools. The system will address standardization at different
layers while providing a single easily accessible web-based interface to store, manage, retrieve,
and analyze data.

27

Chapter 3: Integrated Biomechanics Informatics System (IBIS) Based On
Medical Imaging Informatics Core Principles
This chapter will introduce each component of the Integrated Biomechanics Informatics
System and their core elements. I will go through the system workflow at each step and discuss
the key components involved and the advantages provided that improve the original workflow.
Overall, this will provide a view of the entire system architecture and the flow of data from
collection to review/analysis.
3.1 - System Architecture and Workflow
The IBIS system creates a more efficient and streamlined workflow for biomechanics
research studies by making strong improvements in many key workflow steps. The system
architecture is built to provide a stable, easy-to-access platform that can be utilized at different
locations and also different parts of the research/review process. Figure 3.1 displays the system
architecture and detailed system workflow. In the next few sections, the detailed workflow and
system components will be described.

28

Figure 3.1: IBIS architecture and workflow diagram showing the steps starting with data acquisition
and processing. In the next step the data is parsed and uploaded into the informatics system, which
is in charge of displaying the results onto a web-based interface for analysis and review.
3.1.1 - Data Collection
Data collection for sports performance and injury prevention research (Figure 3.1, Number
1) is an area where an imaging informatics-based system can provide many advantages and also
improve the current data workflow. An informatics system with support tools help streamline the
workflow by assisting in areas such as data standardization, preprocessing, data upload and more.
Advances introduced by the informatics system allows for standardization of raw, preprocessed,
and processed data for analysis collaborating research groups. The informatics system is also built
to support a variety of file formats and integrate them into one centralized location for streamlined
access.
Data Types/Format

29

One of the most important aspects of data collection in the field of sports data and
biomechanics research is to recognize that there are many different types of data collected. This
data can vary on a number of levels, starting with the file type (video files, force curve images,
calculated variables, etc). Then for a specific file type such as video there are different formats
used by different research groups including mp4, mov, and more. Other ways in which video data
can also vary include the frame rate and the zoom.
In the IBIS one of the goals is to standardize the types of data collected by collaborating
research groups so that the data, once centralized, can be compared and analyzed. However, some
sports research labs may prefer to still collect additional data in other formats based on experience
and tradition for data warehousing purposes. The IBIS contains the functionality to accept and
store a variety of data types: those that will be shared across collaborating institutions and those
that will be stored privately. The types and formats of data will also differ based on the nature of
the sport, the research project, and also the devices available at the collection sites.
For the USATF project which includes the long jump and triple jump events, the data types
will include: video data (mp4, mov), force curves (jpg, png), strobe overlay images (jpg, png), and
raw/calculated variables which will be stored in the MySQL database along with the baseline and
collection information.
In the PAC12 project for long distance running, data types will include video data (mp4,
mov) and raw/calculated variables consisting of force/acceleration data which will be stored in the
MySQL database along with the baseline and collection information.
For different collections in biomechanics, there will be parallels between the type and
format of data collected, but also differences depending on the exact nature of the sport or event.

30

The IBIS is built to standardize data coming from different collaborating institutions, but at the
same time it still allows for the flexibility of providing compatibility with a variety of data types
and formats as needed by varying research projects.
3.1.2 - Data Preprocessing with Support Tools
An extremely important step in the biomechanics research workflow is the preprocessing
of data from the collection. The data needs to be preprocessed so that the data can be cleaned up
and synced before the advanced postprocessing algorithms can be run. The preprocessing step is
usually done when the data is brought back to the sports biomechanics lab (shown in Figure 3.1,
Number 2), however, the IBIS will include these support tools integrated into the system so that
preprocessing can be done remotely with internet access. The aim is to keep the preprocessing step
flexible within the workflow because sometimes it may be performed when internet access is not
readily available, such as when traveling back from data collection sites on the car, train, plane,
etc. This flexibility and the powerful preprocessing support tools will help biomechanics
researchers quickly and efficiently clean the data and prepare for postprocessing.
Video Cropping
The first step in the preprocessing workflow is to crop the collected video to only include
events of interest in the athlete trial. During collection, in order to ensure nothing important is
missed, the video recording is always turned on before the athlete performs the movement and
turned off for a period of time after the movement. This is very useful to effectively make sure no
data is missed, however, it adds a lot of time to the video file that is otherwise not needed for
analysis. Working together with the USC biomechanics lab, a video cropping algorithm was
prototyped in MATLAB that will be integrated into the IBIS web user interface. The algorithm

31

reads in the RGB pixel values of every frame in the video starting with the frame before the athlete
enters the frame of view. When the athlete enters the frame of view and exits the frame of view,
there will be a large change in the RGB pixel values. By thresholding these pixel value changes,
the algorithm will be able to determine when the athlete enters and exits the frame of view, which
is the range of frames needed for further processing. This step saves an extremely large amount of
processing time in later pre/postprocessing steps. As an example, after the camera is started in a
trial, the athlete may be stretching, mentally preparing, making last minute adjustments, etc, before
starting to move. This extra time could range from a few seconds to a minute or so, and with the
high frame rate of these collection cameras, this translates to hundreds or thousands of extra
frames. By cutting out these extra frames, it greatly reduces the amount of processing time needed
for following steps, highly increasing research workflow efficiency.
Automatic Foot Contact Detection
Another essential preprocessing step the biomechanics research workflow of track and field
studies is identifying the time of foot contact during a recorded trial. As mentioned in previous
section, the time of foot contact provides a reference that allows for further processing to create
postprocessed data such as force vector overlay videos and also helps sync data coming from
different sources, for example, video, force, and acceleration data. In the current workflow, the
foot detection is done manually in the preprocessing code by a biomechanics expert who visually
confirms the exact frame where foot contact occurs with the force plate. This is an issue because
sometimes these videos have hundreds to thousands of frames and going through each frame one-
by-one before reaching foot contact takes an extremely large amount of time. A code was
developed to greatly reduce the preprocessing time by automatically detecting the frames near the
frame of foot contact.

32

Similar to the video cropping technique used in the section above, the algorithm for the
foot detection uses the RGB pixel values of each of the frames. First, the area of interest is limited
to a bounding box around the force plate where the foot strike is. Then, the RGB pixel values
inside the bounding box are recorded for each of the frames. The values of the original frame are
used as a reference and compared with the following frames using the Manhattan distance. As
shown in Figure 3.2, when the foot first enters the frame of view, there will be the peak
representing a large change. By using this method, the algorithm helps skip all of the frames before
this event. By starting at the frame when the foot enters the frame of view, the biomechanics
researcher only has to scroll through a much smaller number of frames to get to the foot contact
frame. Instead of trying to have the algorithm output the exact frame of contact, based on lab
expertise and also proven methods quality control checks, the gold standard for foot detection is
the expertise of the biomechanics researcher. This method retains the standard of accuracy from
the original workflow while reducing the processing time immensely by removing the need to
manually click through countless frames one-by-one. Once the frame number is identified, it can
be used for further postprocessing steps and syncing the data from different sources.

Figure 3.2: Example of the automatic foot detection algorithm prototype written in MATLAB. The
code will detect the frame numbers (x-axis) where there is a large difference in pixel values (y-axis)
compared to the first frame used as a reference.

33

The first prototype for the automatic foot detection tool will be written in MATLAB and
can be used locally on any computer or laptop either at the collection site, during travel, or back
in the research lab. The next step is to integrate the tool into the IBIS system on the web user
interface so that the processing can also be done remotely anywhere with stable internet access.
There are advantages to both methods as main practice by biomechanics research labs is to save
time by accomplishing a lot of the preprocessing/postprocessing steps during long travel times
where internet may not be accessible, so having the tool available locally and through the web will
be extremely helpful.
Time Syncing/Key Event Identification
The method for acquiring the frame of foot contact in the videos is an extremely important
preprocessing step because the point of contact acts a key event that helps to synchronize data
from multiple sources. The frame of contact helps the biomechanics researcher time sync the video
data along with the force and acceleration data. This allows for further processing and deeper
analysis by creating advanced postprocessed data such as force vector overlay videos and more. A
challenge is that the data comes from different devices. For example, video data is collected from
cameras which are turned on and off for each trial and foot contact is determined visually from the
video frames. For the force data is collected by the force plates over time and are in the form of a
waveform with peaks representing recorded forces. In order to address these challenges, during
the postprocessing workflow, the biomechanics researcher, with the help of the frame/time of foot
contact, can identify this key event and link it with the corresponding recording in the force data.
Once the video and the force data are synced together, biomechanics lab can use the postprocessing
code to create corresponding force vectors and overlay them on each of the frames of the video.

34

This creates a new video that will have real-time force vectors displayed on the video as it is
played.
The data preprocessing support tools help streamline the biomechanics research workflow
by greatly decreasing the amount of processing time and effort needed by the researcher while
maintaining the same level of accuracy. The research workflow will be made even more efficient
and flexible once these support tools are integrated into the web-based IBIS platform allowing
users to perform preprocessing both locally in the lab and also remotely with internet access.
The IBIS is a robust platform that is modular and can support the integration of support
tools. In addition to the preprocessing tools discussed above, any additional tools that are
developed can also be integrated into the system. This potential for continuous expansion and
development removes the limitations associated with one-off standalone systems.
3.1.3 - Data Model/Organization
In order to standardize the method of organizing data across all collaborating research
groups, one single data model is adopted across all institutions. The data model is designed after
the patient-centric structure from the ePR model discussed in Chapter 2 so that the specific data
needed from each trial can be located in a straightforward method. For the example shown in
Figure 3.3, the data will be organized with the athlete at the highest level (main folder) similar to
the patient folder in the ePR model. Then for each the athlete there will be their collections at the
next level (subfolders inside the main folder). Lastly, each collection will have different trials
(subfolders inside the collection folder) and for each of the trials there will be files associated with
that trial (files inside each trial folder).

35

Figure 3.3: Example standardized data model for sports performance and injury prevention research.
An athlete can have multiple collections at different field sites. For each of these collections there
may be different numbers of trials for each athlete and for each collection event. Then for each trial
there are a variety of file types including raw and processed data. These data types vary depending
on the collection site and also the research group.
By using a common data model, research groups will be able to share data easily and locate
the data they need. This solves the issue of confusion between different groups using different
methods of folder structure/data organization that they each may not be familiar with. Standardized
organization of the data with the data model also plays a role during the quality check when
uploading through the gateway/parser. When using the auto-uploader tool, if they data is not
organized correctly, it will not be uploaded into the server (this will be discussed in further detail
in the Auto-uploader/Parser section).
3.1.4 - Data Upload
Once the data has been collected and preprocessed it can be uploaded through the gateway
to the database on the IBIS system which is hosted on servers in the USC EGG server room (shown
in Figure 3.1, Number 3 and 4). The IBIS system also includes the gateway, web server, and the
data-mining module. The upload occurs through the gateway/parser and the data that is uploaded

36

is in a standardized format agreed upon by collaborating research groups. This allows for cleaner
storage of the data and efficient querying/review of the data.
3.1.4.1 - Web Server/Web-Based User Interface
The web server is the piece that hosts the web-based user interface so that the system can
be reached online (Figure 3.1, Number 5). The web server is run with LAMP and Apache 2 on a
server in the USC EGG data center. The Apache 2 web server is configured to communicate with
the MySQL database which stores the metadata for the files and also the recorded raw data and
processed data. This component also connects with the database which stores the multimedia data
including the video data, force data, accelerometer data and more. The web site is built with HTML
and Javascript client side and with PHP and MySQL server side. The HTML and Javascript allows
us to display static and also dynamic elements to the user on their web browser when they reach
and interact with the page. The PHP is responsible for communicating with the server and
retrieving information stored in the MySQL tables. The user interface connects different
collaborating research groups to one centralized location.
The web-based graphical user interface was designed with input from different
biomechanics research groups to be as streamlined and easy to use as possible. First, the user has
to log into a log-in page where they must enter their authenticated username and password. Once
logged-in, they will reach the main Subject List page which shown the list of athletes assigned to
that user (shown in Figure 3.4 below). After the user selects an athlete, they will be taken to that
athlete's main page where they can select from different collections to upload and view data. On
this main page user can also access the built-in tools such as the notepad, data viewer, mass auto-
uploader, etc. (which will be discussed in later sections).

37

Figure 3.4: The web-based graphical user interface has a clean and streamlined design. Once the
user has logged-in they are taken to an athlete list shown in the top screenshot. After the user selects
an athlete they are taken to the athlete’s page shown in the bottom screenshot. On this page, the user
can select and enter different collections to upload/view data, access built-in tools, or use the mass
auto-uploader.
Data can be uploaded to the database using the uploader that is built into the website GUI.
These upload modules shown at the bottom of the web GUI in Figure 3.4, are custom built and can
designed based on nature of the collaborative research project and the agreed upon shared data
model. The methods of upload fit the needs at varying data collections. Data can be uploaded
inside the modules representing “subfolders” inside an athlete’s collection and can be created to
function as subfolders representing a dataset for a specific trial or specific data types. Once these
modules are created, the data can be uploaded inside and displayed on the web GUI in the
standardized organization agreed upon by collaborating research groups.

38

After discussing with various biomechanics research labs, some research groups would still
like an area to upload data in the native method of organization or another method for warehousing
purposes. They have the option of using a secure 3
rd
party file transfer platform such as Filezilla
to securely upload entire folders. In the web GUI I built an additional component of the interface
that will allows them to view these uploaded folders and its subfolders, which will retain their
organization and files (also shown in Figure 3.5).

Figure 3.5: Data can be uploaded into the database in the standardized method of organization based
on an agreed upon data model. In the Figure this is the page for Athlete: TEST1, Collection: Chula
Vista – 3/17/17, and on this page users can upload and view files associated with the Trials at this
collection. The exact organization can be customized with the custom upload modules located at the
bottom. Under the “Uploaded Folders” section in the middle, research groups can still upload their
own folders and the same folder structure will be displayed on the web interface for their research
group only. These folders and their subdirectories can be expanded and shrunk as needed to view or
download files.
To further increase the functionality of the data upload component, a general uploader is
also available at any time that can upload data into any athlete’s collection without selecting and
entering that specific athlete’s collection module. This uploader, shown in Figure 3.6 below,
includes a text box for the user to name the file and toggles to select which athlete and collection
to upload the data under.

39

Figure 3.6: Data can also be uploaded at any time to any specific athlete’s collection by using the
general uploader and the appropriate built-in toggle selections.
3.1.4.2 - Auto Uploader/Parser
The organization of the stored data is key to the centralization and streamlined access of
the data for multiple collaborating institutions. When uploading into the server, users can create
upload modules (shown in the bottom of Figure 3.5) for each collection and upload the correct
files into each one, following the agreed upon data model by all collaborators. This type of
organization in the server is designed to mirror the standardized data model of folder/subfolder/file
organization used by the research groups to ensure that once the data is in the server it is easily
located.
One tool developed that is built into the gateway/parser of the server is an auto-uploader
that streamlines the upload process allowing a user to upload an entire folder, which contains
subfolders and data files, and the files will be automatically saved in the standardized organization
on the server. The auto-uploader will first read and parse the folder and subfolder structure to make
sure that the correct standardized organizational structure is used. Once this quality check is
performed and cleared, the information parsed from the folders and subfolders inform the system
of the specific athlete/collection/trial under which the files need to be uploaded. Once the system

40

has this information, each of the files in the subfolders of the main uploaded folder will be
automatically uploaded into the server with the correct date model organizational structure
(displayed in Figure 3.7 below).

Figure 3.7: The gateway/parser in the IBIS server contains a built-in auto-uploader that can take in
an entire folder and parse its files into the correct locations in the standardized data model on the
server. The gateway/parser will read the folder/subfolder organizational structure of uploaded
folders and only allow upload for folders with the correct structure. Using the information from the
gateway/parser, the auto-uploader uploads each of the files to the correct location on the IBIS server,
shown by the grey arrows.
3.1.4.3 – Remote Access
Key elements I built into the informatics system were centralization and accessibility. The
data is housed in a central area and that it would be easily accessible through a graphical user
interface. The web-based user interface can be accessed through the Google Chrome, Firefox, and
Internet Explorer web browsers. As long as the user has connection to the internet, they are able
to access the informatics system. The user interface allows for streamlined and user-friendly access
to one centralized location where all authorized users can complete the tasks needed whether it is

41

uploading, managing, or reviewing data. Research groups now have the flexibility to upload the
data during the collection at the field site while optionally backing up data on a hard drive for a
physical copy to have one hand. However, sometimes at the collection site when working with
multiple coaches and athletes, there may not be enough time in between trials to upload everything
right away. In this case, research groups also have the option to upload the data to the informatics
system when they return to the lab or a hotel room (if collection site was in another state, country,
etc) as long as there is an internet connection.

Figure 3.8: Centralization of the data in one location allows for more efficient access from different
users. The web server in the integrated informatics system hosts the web-based user interface for
users to upload data, manage data, review the data, and more. The different users including
collaborating research groups, coaches, athletes can reach the informatics system remotely from any
location with internet access through a web browser.
The web-based system now gives research groups much more flexibility during the
research collection process in regards to data collection and storage. It also allows collaborating
research groups, coaches, and athletes themselves to view and download the data wherever they
are as long as an internet connection is present. For more complex review and analysis of the data,
the web interface also includes a built-in streamlined data viewer module that allows for
IBIS SYSTEM

42

simultaneous viewing of video and data. On the web interface users are also able to view the data
using the streamlined data viewer module and access data analysis tools which will be discussed
in detail in later sections.
3.1.4.4 - Security/User Authentication
Another core principle I built into the IBIS system is security and protection of the data
that is stored. The user interface that is hosted on the web server is in the form of a website that
contains a log-in page for access. Researchers, coaches, and athletes are assigned user accounts
and they can create their own username and password. To log into the informatics system users
must use these authenticated user accounts. One essential aspect to incorporate into the informatics
system was security. The user authentication process mentioned above was the first step of
development in this area. The next step was to include a firewall that would protect the system
server from online cyber-attacks and undesired web traffic. This is accomplished by hosting our
informatics system and its web server inside the USC EGG data center. In doing so, the server is
protected by the USC ITS firewall which allows us to monitor for any issues. The USC firewall
put in place will track any suspicious activity or traffic. The ITS will provide notifications of any
attempted suspicious activity and also any new potential cyber-attacks that are circulating around
the web.
3.1.4.5 - Privacy/User Access
Related to security is the core principle of privacy and controlling user access that is built
into the system. As mentioned in the section above, once the data is uploaded in the database only
authorized users can log in to access the data. This is the first step in controlling access to the data,
however, further access control in greater detail is needed. Once logged into the system a user will

43

only be able view the data that they were assigned to, for example a coach would only be able to
see the data for his athletes (shown in Figure 3.9 below). An athlete who logs into the system can
only view his or her own data.

Figure 3.9: Web graphical user interface displaying the anonymized athlete list a user would see
after logging in. Different user access controls can be set for a variety of users such as admins,
coaches, athletes, researchers, etc. (Top) shows limited user access to only specific athletes and
(Bottom) shows admin access to entire list.
This second layer of data privacy control is also very useful when it comes to multi-
institutional collaboration. For example, if there are multiple institutions collaborating on a
research project and they each have data for their own athletes, there may be data that they would
like to share with the other groups and data that they may prefer to keep private. On the IBIS it is
possible to create a version of an athlete who is visible on the Subject List for all the institutions
and the data uploaded can be viewed and downloaded by all. Another version of the athlete can be
created by their institution and this athlete version would only be accessible/visible on their

44

institution’s list. Any data uploaded for this version of the athlete would be private and not shared
with the rest of the collaborating groups.
The log-in authentication gives the informatics system a layer of security and privacy.
Controlling the user access further adds another layer of protection. An additional layer of
protecting data privacy is implemented with the anonymization of the data that is discussed in a
later section.
3.1.5 - Storage/Management
As mentioned earlier, once data has been preprocessed/processed and standardized it is
uploaded securely to our data center within the USC ITS firewall. Files such as video and force
curves are uploaded in a similar structure to the data model in Figure 3.3 that is agreed upon by all
collaborating research groups. Numerical calculated data for the trials, including both discreet and
continuous data will be uploaded into the MySQL database management system (Figure 3.1,
Number 4). The tables will store not only the data itself, but also associated metadata and file
locations. When queries are run in the GUI to view or download the data, the file locations are
retrieved from the MySQL tables. The table also helps maintain the data model mentioned above,
tracking the organization of the data, for example, which trials belong to which collection. All of
these elements can be easily managed by the system administrator on a streamlined interface.
3.1.5.1 - Data Anonymization
Building upon the security/privacy core elements mentioned above, once the data is in the
system and being managed or downloaded, it has to be anonymized to protect to athlete privacy.
For projects involving multi-institutional collaboration, a list of randomly generated IDs is created
and sent to all groups. Each institution assigns a random ID to each of their athletes. This is to

45

protect the privacy of the data when it is shared in a centralized location in order for all groups to
collaborate and review/analyze the data. Only the athlete themselves, their coach, and the research
group they are working with at the institution will know their ID number on the server. The data
anonymization is an additional layer that protects the privacy and the security of the data on top of
the pre-existing user authentication and user access controls mentioned in the previous sections on
security and privacy.
3.1.5.2 – System Admin Control Panel
The IBIS platform contains a built-in control panel for the system administrator. This
streamlined interface allows the system admin to quickly manage users, data, and organization for
the server. This is shown in Figure 3.10 below, where the system admin can create new subject
users while assigning them to the correct coach and institution that is necessary for the access
controls discussed in the next section. The admin can also use the interface to create new
collections and update new information or correct items that were entered incorrectly without
having to access the backend code of the system. These same actions can be performed for any
multimedia files and associated metadata that are uploaded onto the server such as reassigning the
file to another subject/collection, renaming the file, and/or replacing the file.

46

Figure 3.10: Built-in control panel for the system administrator that allows the admin to quickly
update subject users, collections, files, and associated metadata with streamlined toggles/fields.
3.1.5.3 - User Access Controls/Permissions
Once the anonymized data has been uploaded into the IBIS server, authorized users are
able to log in to view the data. However, as mentioned in the Security/Privacy sections, the system
needs to control exactly which user is allowed access to specific athletes. For example, a coach
would be given access to view all of his athletes’ data while athletes themselves would only be
given access to their own data. This system also allows collaborating research groups to control
how much and which data they would like to share. In the informatics system there is a shared
space where all data uploaded is shared with all the collaborating groups. Each group still has the
option of uploading to a space that is private and only they can access.
These user access controls can be controlled in further detail on the informatics system by
assigning what actions the user can complete on the server. Complete access gives a user the
capability to upload, download, view, rename, and delete data. However, in certain situations such

47

as accessing shared data from multiple institutions, research groups did not want to users to be
able to rename and delete data as it could lead to confusion. These specific access rights for users
can be assigned such that certain users can only upload/download/view data, while an admin user
is allowed to rename and delete files as well.
3.1.5.4 - Data/Metadata Storage
The IBIS system incorporates many different forms of multimedia data from different
institutions and stores it into the server. It also stores any numerical data and variables that are
recorded or calculated. For example, the force data and acceleration data come in the form of .dat
files or .txt files depending on the collection device and set up. Then the data is usually input into
a program such as MATLAB or R for analysis. The IBIS system allows collaborating users from
different research institutions to upload the data first into a centralized MySQL database. The
centralized database is a single location where the data will be standardized in MySQL and easily
accessible by all users.
The centralized database is essential because it stores not only the raw data but also any
processed data. Different research groups may choose to view/process/analyze the data in different
programs such as MATLAB and R. Each of programs that the different labs choose to use are an
essential part of their research workflow that has been built upon year after year and our IBIS
system is built to be flexible and accommodate the needs of the collaborating institutions while
still providing standardization and centralization. The MySQL database, which stores all the
variable data and metadata (discussed in the following paragraph), will be accessible through
connections in both the MATLAB and R programs. This way data can be stored/retrieved and then

48

viewed/analyzed without the user needing to change from a program they are familiar with using
to downloading and learning to use a completely new one.
There will be accompanying metadata, which is information about the data. For example,
the format of the data, the source of the data, how it was obtained, etc. Similar to the metadata
associated with medical images, the metadata for sports video, force files, and more are extremely
important because they allow users to not only see more details about the data format, but also
how the data was collected (devices used, settings, environmental factors during the collection,
etc). This is especially important in collaborative projects between multiple institutions because
the set-ups and settings may differ for each institution.
The metadata will be stored along with the calculated values such as force and acceleration
in the MySQL database. For each trial that is stored in the database, the information for that trial
such as the collection date, the location of the collection, the devices used, the athlete’s body
weight, and many more key factors. In the PAC12 case study (will be discussed in detail in the
Case Study section), the metadata is extremely important because some institutions collect data on
the treadmill and each of those trials will consist of many steps, while for other institutions each
trial is each step on the force plate. Once the data is standardized (discussed in the Data
Standardization section), the treadmill data will be sliced into each step to match the format of the
per-step data, however it is essential to record the time points where the treadmill data was split
up and also the indices for each of the resulting pieces. This metadata will be stored along with the
trial data itself, so that if a user needs to access it for analysis/reference, it will be readily available
in the database.
3.2 - Knowledge Discovery and Data Analysis Tools

49

Once the data is securely uploaded into the server and users can efficiently manage the
storage and access to the data, the next step is to review the data and analyze it. The graphical user
interface built into the IBIS system allows users to quickly locate the data of interest, download
the data, or open the data for viewing with the built-in tools. Tools on the user interface give users
the ability to view videos, look at force curves, and more without the need to download or install
different software as would be needed on a personal laptop or computer. A streamlined data
viewing module has also been developed which has the capability to efficiently display data from
multiple multimedia sources simultaneously for advanced data analysis. Data-mining tools
including statistical analysis tools are integrated into the system. The IBIS provides the user
(researchers, coaches, athletes) with these tools to perform advanced data review and analysis for
knowledge discovery and decision support (Figure 3.1, Number 6). These built-in decision support
tools each play important roles in the overall eAR system and because of the robust nature of the
IBIS, the modular tools and any additional new tools can be easily incorporated into the system.
3.2.1 - Data-Viewer Application
A web-based graphical user interface was created to streamline the management and
retrieval of data. As mentioned in previous sections, the data can be located on the system and
downloaded locally onto the user’s local computer if needed. The data can also be opened directly
on the web interface with built-in data viewing tools. There is no need to download and install a
specific program to run in order to open the data files. As shown in Figure 3.11, the user can open
a video in a built-in video viewer on a new tab by simply clicking the “View” button for the desired
video. Likewise, by clicking the “View” button for a Force Curve image, will open the image
within an image viewer on a new tab. These tabs are very flexible and can be moved around and
enlarged or minimized depending on user preference. This functionality in the system saves a large

50

amount of time that is usually spent on making sure specific programs are available to open these
kinds of files and opening these programs/waiting for them to load. The functionality also prevents
the user from having to open multiple programs to view different file types such as video and
images, potentially overloading and slowing down their local computer or laptop.

Figure 3.11: By selecting the “View” button for the desired data on the graphical user interface,
users can open and view the files using viewers that are integrated into the IBIS system. On the right
side of the figure is a built-in video viewer. On the left side is a built-in image viewer. These viewers
can be adjusted and moved around depending on user preference. By opening these files for review
all in one location on the web interface, there is no need for opening and running multiple programs.
3.2.1.1 - Synchronized Multimedia Data Analysis
As mentioned in the section above, selected data can be opened and viewed with built-in
viewers on the integrated informatics system. For more advanced data analysis and review, I
built a streamlined data viewing module that allows for streamlined simultaneous review of data.
This viewer is designed so that data can be located and retrieved very quickly and where exactly
the data will be positioned on the viewing screen can be customizable.

51

The developed web-based data viewing dashboard is shown in Figure 3.12. On the top left
of the data viewer it will show the deidentified athlete ID and also the coach ID. Under, along the
left side, the athlete’s different collections will also be listed. Once a collection is selected, the data
from that collection will appear in a column underneath. The column of data is organized by file
type, for example, raw video files, overlay video files, force curves, etc. And then under each data
type there will be a list of the corresponding data of that type for that collection.
The four blocks are viewing modules that can be selected for viewing the data. A user can
first click the blocks that they want to use and then select the data they want to view and it will be
retrieved and loaded into the selected block. With the different data viewing blocks, the users can
choose to view both video files simultaneously with the force curve data, all videos, all force
curves, etc. This customizable setup allows researchers and coaches to place what data they want
to view at their preferred positions and streamlines the data analysis/review process so feedback
can be given to athletes as effectively and efficiently as possible. For example, in Figure 3.12,
there are 2 videos opened adjacent to the 2 associated force curve images, but if preferred, the
viewer can be chosen to display 4 videos or 4 force curve images at once.

52

Figure 3.12: The data-viewing dashboard user interface will include the athlete deidentified ID and
the coach name with their collections and collected data for that trial on the left. On the right side,
the bulk of the data viewer is made up of four viewing modules that can display raw video, force
vector overlay video, force curves, calculations, etc.
3.2.1.2 - Customizable Layouts
The data viewing module is streamlined so that the user can quickly select the athlete that
they would like to view data for. Then once the athlete is selected, the user can easily locate and
choose the different data that they would like to review in the data viewing sections. The specific
placement of the desired data in each of the squares is fully controlled by the user. The data viewing
module has another advanced level of customization which allows the adjustment of the viewing
layout. Shown in Figure 3.13, there are some data types such as the vector overlay strobe image
with a very specific shape and may not fit the more even block shapes on the starting template
layout. And so, the user has options to use from a variety of different layouts. For example, the top
blocks can be changed into two elongated rectangular blocks which will better fit the longer strobe

53

images. This functionality gives the users countless options and combinations of data view such
as viewing strobe overlay images, force vector overlay video, and force curves all at the same time
on one viewing dashboard.

Figure 3.13: The data-viewing dashboard user interface also has the functionality of displaying
different layout templates allowing for even more advanced customization and data review/analysis.
In order to effectively display different kinds of data of varying shapes, the data viewing blocks in
the module can be interchanged into different combinations. For example, in the figure above, this
template chosen has elongated rectangular data viewing blocks (purple and red rectangles) to better
fit the specialized elongate shape of strobe motion images, while the bottom two blocks (white and
green) can be simultaneously used to view force vector overlay videos or force curve images.
3.2.2 - Data-Mining and Decision Support
Once all the data has been stored into the IBIS, the next essential task is to view and analyze
the data for knowledge discovery and decision support. The data-viewing module, shown in the
previous section, allows the user to view different forms of multimedia and compare them in
customizable layouts. The IBIS system also includes built-in tools to allow users to review
collected numerical values and also any calculated variable from the collections. This may include

54

variable such as raw force measures or acceleration data recorded during the trials. Some key tools
built into the system include a web-based interface that allows the users to track an athlete data
longitudinally over time and across studies for different athletes. The web-interface also includes
an integrated statistical analysis tool allowing the user to perform real time filtering and data
analysis. These tools used in conjunction with the multimedia data viewing module allow for
deeper analysis and knowledge discovery leading to decision support.
3.2.2.1 - Integrated Statistical Analysis Tools
The statistical analysis tool is built using the R Shiny application which provides the
framework for creating clean streamlined user interfaces like the dashboard shown in Figure 3.14.
The R Shiny application is written in the R language which is widely used for advanced statistical
analysis and visualization in many fields of study. R Shiny is integrated into the IBIS server so
that it can be accessed as a web application similar to the rest of the system. Users must log in with
their authenticated credentials to access the application remotely from anywhere with solid internet
access. Users do not need to download the R program or any other 3
rd
party statistical analysis
software to perform any analysis/review of the data. The integrated R statistical analysis dashboard
includes filters, modules, identifiers that can be used for tracking over time and analysis across
multiple athlete subjects.
Longitudinal Tracking
At the top of the dashboard shown in Figure 3.14, users can select the event they want to
review and filter by sex/name of the athlete. In the middle section, the user can select what
variables they would like to analyze together. Users can track variables longitudinally over time
or have the option to look at any combinations of variables by selecting them from the dropdown

55

lists on the interface. Once the variables are selected, they are instantly plotted on a graphical
interface right underneath the filters. In the example in Figure 3.12, the user has selected to look
at data collected during the Triple Jump (TJ) trials for an athlete (deidentified for privacy
purposes). The variables selected are date on the x-axis and post impact duration on the y-axis to
track this variable over time. From the dropdown list shown in the figure, the user can select any
variables they want to look at including velocity change, impact duration, explosive strength index,
etc, and the graph below will immediately update accordingly.

Figure 3.14: This figure displays the user interface for the R Shiny integrated statistical analysis
tool. The statistical tool and user interface are integrated into the web-based IBIS platform and can
be accessed remotely from a web browser and internet access. Using the dashboard interface, users
can filter by event, sex, athlete, and select variables they would like to view on the graphical chart
module. The example in the figure shows an athlete’s post impact duration collected during trials
over time. From the dropdown there are many other possibilities to choose from including velocity
change, impact duration, explosive strength index, and countless more. (Athlete has been
deidentified for privacy purposes).
Cross-Study Analysis

56

From the previous section I showed how the integrated statistical analysis tool can be used
to track different raw and collected variables longitudinally over time. The advanced statistical
analysis interface also allows the user to add a third variable to simultaneously review in addition
to those selected for the x-axis and the y-axis. This is represented by the color identifier module,
the blue box shown in Figure 3.15. This functionality allows the user to perform advanced analytics
such as cross-study analysis of multiple athletes at the same time. The example in Figure 3.15,
displays the selected of variables “peak force” and “jump distance” plotted on the graphical
interface, and the color identifier is set to “athlete”. This will change the color of the plotted points
for each athlete; the colors of each athlete’s data points will be displayed in the legend located in
the bottom right corner (deidentified for privacy purposes). This allows the user to analyze the
relationships between different variables selected and also view these correlations for multiple
athletes simultaneously.

57

Figure 3.15: The R Shiny integrated statistical tool can also be used for cross-study analysis over
multiple subjects as well. This figure displays the user dashboard of the integrated statistical analysis
tool. The user interface includes tools to filter by event, sex, athlete, and select which variables to
display on the graphic interface. The additional color identifier helps add a third variable to the
analysis. In this example, it displays the variables peak force and jump distance across multiple
athletes, represented by the color identifier (athlete names in the legend are deidentified for privacy
purposes).
The integrated statistical analysis tools provide an easy-to-access, streamlined,
straightforward platform for all authorized users to remotely review and analyze the data. The
platform is used by biomechanics researchers for deeper analysis, data mining, and knowledge
discovery to provide high-level feedback to coaches and athletes.
Similar to an electronic patient record the statistical analysis tool has a functionality where
a user can highlight a bullet point of interest and it will display a list of links of the available data
associated with that trial (data point). Effectively, this feature acts as a timeline where data points
can be sorted by date, trials of interest are selected, and the data for those trials readily available

58

for review. Coaches and athletes can use the platform at training facilities for decision support by
reviewing any past trials to acquire insight in how to prevent injury or improve performance.
3.2.3 - Notes/Annotations
Tools for streamlined communication and were designed to address the need for
streamlined collaboration between multiple groups. On the statistical analysis tool there is a built-
in chat and email functionality. Researchers can quickly send emails or record any notes in the
chat while working on the statistical analysis user interface. The annotation and collaboration
module was also built into the web-based system so that there would be a location for collaboration
that all authorized users could access. These solutions solve the issue of having to search through
email threads and allowing users to quickly and effectively communicate with one another. Users
that are logged in remotely to the IBIS system will also have access to this module. In this module
(shown in Figure 3.16), the user has the freedom to type in freehand any notes or annotations they
would like. In this method users are not restricted to using a form or a limited response system
where it may be difficult to accurately convey their ideas.

59

Figure 3.16: The annotation and collaboration module is built into the web-based informatics system
and functions similar to a custom notepad. Users are able to record and save their notes and
annotations on the system for use later. The module also functions as a centralized location for
communication between involved research groups. The example in the figure shows sample
communication and potential knowledge discovery dialogue between researchers, athletes, and
coaches.
Researchers from different institutions are able to add any new findings or thoughts onto
the module and save it. Once this is done so, the notes will be viewable by any authorized users
from collaborating institutions. Athletes are also able to use this module to communicate with their
coaches, who may not be at the same location. On the system, all their data will be loaded and
viewable. The athlete will be able to view their data and any videos from all trials. When they
recall any key events that may have caused pain or any details that may have affected performance,
the athletes can record this onto the module. Once this information is there, the coach can also
rewatch the videos for those trials with the information from the athlete in mind. The coach can
then provide any feedback and thoughts back to the athlete on the same module or communicate
with the biomechanics lab for their feedback on the key events. This module allows for much more

60

efficient and deeper discussion/analysis than conventional methods of trying to communicate
through multiple email threads, text messages, sending files back and forth or setting access
privileges on multiple platforms, etc. The module is a centralized location which combines the
feedback/experience from the athlete, the expertise/experience from the coach, and also data
processing/analysis from the biomechanics sports lab.

61

Chapter 4: Use Cases in Biomechanics Research
In this chapter I will go over the use cases for the integrated biomechanics informatics
system (IBIS). The use cases will include going through the biomechanics research workflow and
the specific needs of an informatics system from data collection to knowledge discovery and
decision support. Projects for the use cases will include exploring injury prevention and
performance enhancement for USATF triple jump and long jump and a PAC12 long distance
running stress fracture study.
4.1 - USATF Triple Jump and Long Jump Project
The first use case is a research project between the USC biomechanics lab and the United
States Track and Field (USATF) team. The research workflow encompasses data collections at
training facilities/competition locations, data processing, storage, sharing, analysis and review.
Key steps and limitations in the workflow will be discussed in detail below along with the IBIS
components that will address these limitations.
For the USATF Track and field project, the biomechanics lab at USC is studying
specifically long jump and triple jump events. Biomechanics researchers work closely with
coaches and athletes during training collections and/or competition collections to set up a variety
of devices and hardware. The hardware that can be set up at each site is different. For example, at
the USATF Olympic Training Center in Chula Vista there are both force plates and video cameras
set up, while at other sites for meets such as the Des Moines site and the Sacramento site there are
no force plates, and only video data is collected. The data is still continually being collected and
the standardized data types and their formats collected for the USATF Track and Field project are
displayed in Table 4.1. The preprocessing steps such as video cropping and automatic foot

62

detection are currently done in software with extensive input from a manual user. There is a need
for pre-processing tools that minimize the manual user input to streamline the research workflow
and help time sync multimedia more efficiently.
Data Type Data Format
Video (includes different angles/zoom) mp4
Video mov
Force Curve Images jpg
Force Curve Images png
Strobe Overlay Images jpg
Strobe Overlay Images png
Force Values Entries in MySQL database
Discrete/Continuous Values Entries in MySQL database
Calculated Variables Entries in MySQL database

Table 4.1: The table above is a table displaying the different data types/formats collected for the
USATF long jump and triple jump trials.
The USATF research workflow is shown below in Figure 4.2. Data is currently stored
locally on laptop/desktop computers, hard drives, and on cloud-based software such as Google
Drive/Dropbox. Using a mixed variety of devices and software leads to log-in authentication issues
along with the difficulty of sharing the data in the form of physical hardware. There is a need to
create a centralized location for storage that can be remotely accessed.

63

Figure 4.2: The USATF workflow contains many steps and different locations for each step such as
searching for old data in CD archives or hard drives, storing data and sharing data in different
locations, and the need for end users to have specific programs/software installed to review the data.
For security and privacy of the USATF athletes, the data for each athlete is only shared
with that athlete and their associated coaches and research teams. Along with the difficulty of
sharing data stored in various locations, only specific data sets can be shared with each user and
this needs to manually configured on each storage platform. For example, the data stored on
Google Drive needs to have exact user privileges set correctly for each folder and copies of
physical drives need to be made containing only the exact data for sharing. The centralized storage
solution will need to include secure user authentication and user access controls where a system
administrator can efficiently create users and manage which data each user has access to.
Another challenge in the research workflow is the organization and retrieval of data. Each
athlete contains different collections at each site, at each collection site there are many trials

64

recorded, and for each trial there are various forms of multimedia data and variables collected. The
data is stored and shared in various locations and may be organized in different ways, which makes
locating specific multimedia to review time-consuming. Once this data is located, the files are
opened using a variety of third-party software, which not only take time to load/update, but need
to be pre-installed on the device being used. The centralized informatics system used to store the
data will need to include a streamlined graphical user interface for users to quickly locate data that
has been organized. The system will be remotely accessible through a web-based platform so that
it can be accessed by users from any location with internet access. The web-based system needs to
include built-in functionality for viewing the different types of multimedia without the need for
outside software installation. In the USATF research process, there is also a need to review
multiple types of data such as video and force curve images at the same time. Currently, to view
multiple multimedia files simultaneously, it requires the user to open multiple software
applications and resize the windows to fit them on the same screen. The windows need to be resized
to effectively analyze specific types of data such as the elongated rectangular force vector filmstrip,
an essential element in track and field research. A streamlined data-viewer application on the
informatics system would allow the user to retrieve any set of data stored on the server in real-time
and display them simultaneously on a single organized viewing module that is robust enough to
support a variety of multimedia and includes various display template layouts.
The USATF project also involves the analysis of calculated variables. These pre-processed
and post-processed variables are input into 3
rd
party statistical software such as MATLAB and R
to run statistical analysis for knowledge discovery. Although this process may work for an
experienced researcher with programming experience, it represents a challenge for coaches and
athletes who don’t have the experience in coding or the time to learn. There is a need for a method

65

in which all users including USATF athletes and coaches can explore the data with a streamlined
analysis tools without needing to install any software or coding. This tool will be able to perform
real-time queries on the spot on any data set that the user would to analyze or share.
Once the data is shared between researchers, coaches, and athletes, the next step is to
analyze the data and discuss key findings for decision support. The goal is to use the information
and develop strategies to prevent injury and improve performance for the athlete. The
communication between all parties is done in person at collection sites and through email. Every
time there is a key finding, the researchers can send an email to the coaches and athletes, however,
as discussions continue, this results in longer email threads and multiple email threads. The
information becomes increasingly harder to organize and keep track of. There is a need for a
communication tool built into the informatics system that allows for conversation and discussion
on research findings to be saved in a centralized location. All participating users will be able to
quickly access the discussion platform and save their comments and contributions promoting
efficient communication and collaboration.
4.2 - PAC12 Long Distance Running Stress Fracture Study
The second use case is a multi-institutional research study within the PAC12 with 4
participating groups: USC, Colorado, Oregon, Stanford. The project is a collaboration between
athletes, coaches, and biomechanics labs at all 4 institutions. The participating research groups are
exploring stress fracture studies in long distance running. The sections below will discuss the
research workflow (shown in Figure 4.3) with an emphasis on multi-institutional communication
and collaboration and areas where novel tools of an informatics platform provide advantages that
were otherwise not possible.

66

Figure 4.3: The original research workflow of the PAC12 Study for Long-Distance Running shows
the collaboration between the 4 institutions. Once each of the research group collects the data, it is
stored onto a local device and various storage locations (physical drives, Dropbox, etc).
Communication and collaboration are done through micromanaging files on Google Drives and
multiple email threads.
The biomechanics labs at each of the institutions work closely with the track and field team
at each of their respective schools. The data for USC is collected on an outdoor track site and the
data for the other three institutions Colorado, Oregon, and Stanford is collected on an instrumented
treadmill in the research labs. The standardized data types and their formats for the PAC12 project
are shown in Table 4.4. Similar to the use case for the USATF project, the data is stored using
physical drives and 3
rd
party platforms such as Google Drive and Dropbox. The data is also stored
under different organizational methods at each institution. There is a need for a centralized location

67

for storage and set up a standardized data model of organization that will enable streamlined access
to the data for sharing.
Data Type Data Format
Video (includes different angles/zoom) mp4
Video mov
Force Curve Images jpg
Force Curve Images png
Strobe Overlay Images jpg
Strobe Overlay Images png
Force Values Entries in MySQL database
Acceleration Values Entries in MySQL database
Discrete/Continuous Values Entries in MySQL database
Calculated Variables Entries in MySQL database

Table 4.4: The table above is a table displaying the different data types/formats collected for the
PAC12 long distance running stress fracture study.
Once the data is collected, preprocessing and standardization steps are essential in the PAC
12 project because of the differences during data collection for different institutions. The USC
track and field data will be collected with force plates as a series of steps while the other three
institutions will be using instrumented treadmills with a continuous set of data. The force and
acceleration data needs to be standardized across all institutions and currently these algorithms are
run manually by the user, consuming a large amount of time. In order to streamline the process,
tools such as the video cropping and automatic foot detection tools can be used on the data
collected from multiple PAC12 collections to help time sync multimedia more efficiently.
In the PAC12 study, the data is treated similar to that of the USATF project where each
athlete has access to their own data along with their associated coaches and research teams at each
institution. However, for the PAC12 project there is an additional component where subsets of the
data are anonymized and shared between the collaborating research groups for cross-study
analysis. This requires significant micromanagement of managing the exact sets of files to share,

68

making copies of physical drives, and making sure to set the exact user access controls across
different platforms. A centralized location for uploading, storing, and sharing data along with a
streamlined user interface system provides not only user authentication but also the ability to
assign specific access rights according to what datasets will be kept private or shared.
In order for the research teams to share multimedia data and findings, files are shared
through Google Drive or Dropbox and need to be downloaded by each institution and opened with
the appropriate program. This requires each collaborating location to have all the necessary
platforms/software installed and learn how to use them. This involves downloading multiple files,
keeping track of the downloads, and then opening/resizing multiple windows to view multiple data
simultaneously. An integrated informatics platform will allow all users to query any multimedia
file right away and view the file with built-in data viewers. There will also be a need for a
streamlined data-viewing module that allows users to simultaneously review and analysis multiple
multimedia data, removing the need for any additional software.
The key emphasis of the PAC12 study is the analysis of measured force and acceleration
variables across multiple subjects to promote discussion and knowledge discovery. Currently the
analysis is done locally by each institution on the shared data using statistical analysis software
such as R or MATLAB and the results are shared through email threads. Analysis performed on
different software may lead to issues in standardization and further review of the data. Or the
decision for all institutions to agree on one software to use comes with the challenge of requiring
researchers, coaches, and athletes at participating institutions to learn how to use that software.
The goal is to have a built-in statistical analysis tool that is software-less and is web-based that can
be accessed remotely. Users with or without any programming knowledge can use the tool to
analyze the data on the server in real-time using streamlined user interface buttons and filters. In

69

addition to the analysis tool, there is a need for a communication tool or module that allows for
streamlined communication between all users in a centralized location. Users will be able to save
their notes and any questions to share with collaborators while analyzing the data. This tool is best
located on the same platform hosting the data sharing and review to further streamline the research
workflow to promote discussion, knowledge discovery, and decision support for the study.

70

Chapter 5: Results/Evaluation
The IBIS system was evaluated using the two use cases of the USATF project and the
PAC12 project. Significant results were achieved based on key principles of functionality,
efficiency, impact on knowledge discovery and through workflow efficiency comparisons of
performing essential tasks in the biomechanics research process.
5.1 - USATF Triple Jump and Long Jump Project
For the first use case, the IBIS was integrated into the USATF project research workflow.
The new research workflow integrated with the system and support tools was tested at and with
the data collected from research labs, training facilities, and competition locations. All key steps
in the workflow are discussed in detail in each of the following sections with workflow
comparisons as shown in Figure 5.1 below.

Figure 5.1: A comparison between the original USATF research workflow and the new workflow
with the IBIS system. The original workflow contains many steps and different locations for each

71

step such as searching for old data in CD archives or hard drives, storing data and sharing data in
different locations, and the need for end users to have specific programs/software installed to review
the data. The IBIS integrated system significantly improves the workflow by streamlining it through
one centralized platform where all tasks/steps can be accomplished.
5.1.1 - Data Collection/Preprocessing
Data was collected at a variety of training sites and competition locations for the USATF
project. The data included the data formats discussed in the previous use case section for the
USATF project. As multimedia data and numerical values are continuously being collected and
processed for the project, raw, pre-processed, and post-processed data can be directly stored into
IBIS database at any step of the research workflow. Once the data is store in the database it can be
retrieved at any time for further pre/post-processing. The prototype of the video cropping and
automatic foot detection tool are still currently in testing, but is able to greatly reduce the number
of frames that the researcher has to manually check through for cropping and identifying key
events.
5.1.2 - Data Upload/Storage
The server that hosts the web-based IBIS platform is installed in the USC EGG Data Center
to ensure secure connections monitored by USC ITS (Information Technology Services). As
shown in Figure 5.1 users are able to access the web-based platform remotely from any
computer/laptop with internet connection and upload/manage data as needed. The IBIS database
is used to store both raw and processed data in a single centralized location that can be accessed
for data retrieval at any time. Data from past collections can be uploaded from older CD archives
or hard drives and will be readily available for any future review. The data is organized in a
standardized method based on the ePR framework discussed in Chapter 2, with the athlete at the
highest level so that all users can quickly locate the desired athlete subject. The web-based user

72

interface also contains the necessary user authentication protocols. Only authorized users including
coaches, athletes, researchers, and clinicians are able to log into the platform and
upload/manage/review the data. The system has user access controls set by the system
administrator which assigns specific access rights to each user controlling exactly what data the
user can view and also what actions the user can perform on the data. These steps ensure the
security and privacy of the athlete data is protected within the IBIS system. This single centralized
IBIS platform creates a more streamlined workflow that was otherwise not possible and eliminates
the need of multiple storage locations/devices in the original workflow as well as the need to secure
each of the devices with the appropriate security /privacy protection.
5.1.3 - Data Review/Analysis
Once the USATF data was uploaded into the IBIS system. Coaches, athletes, researchers
from various remote location were able upload and download data as needed. In addition to this
new functionality, further advancements are made possible with the web-based user interface. The
IBIS system allows for solid practices, clarity, and speed for all users to easily navigate to the
athlete data they want to review. The functionality for opening and viewing video and image
multimedia files are built into the web-based user interface so there is no need for users to
download, install, or open additional software (shown in Figure 5.2 below). Once logged in, users
can quickly select the athlete and the specific collection for that athlete that they would like to
view data for. Then once inside that collection, all the available data types for the trials will be
listed by category for quick access. The stored data can be viewed and downloaded by all users,
with the system administrator having the access to rename and delete files as well. In the example
shown in the figure below for the USATF project, the user is able to locate the data for a collection
for triple jump (athlete name, date, collection location have been de-anonymized for privacy). The

73

user has opened a Force Time Plot to view along with the Side View Video, and can play the video,
rewind, or open other files as needed for real-time review.

Figure 5.2: The web based graphical user interface allows the user to select the athlete and collection
they would like to view data for. Inside the collection, the user is able to select the specific data of
each type they would like to view. By clicking the view button, new windows are opened as video
player/image viewers instantly, which are all built into the online platform. These windows can be
moved around and resized as needed.
The user interface displayed in Figure 5.2 above shows how video and force data for long
jump or triple jump trials can be opened and displayed without any need of outside software. New
windows will be opened automatically by clicking the view button, which can be moved and
resized to fit the specific needs of each user. This part of the interface allows for the quickest access
to an athlete’s data real-time.
For more detailed analysis reviewing data from multiple trials and of different file formats
simultaneously, the IBIS server has a built-in data viewing module. The data viewer is a
streamlined graphical user interface (shown in Figure 5.3 below) allows users to toggle through

74

all of the athlete’s existing collections and select any videos or force curves they would like to
look at. The selected multimedia files will then be displayed real-time on the viewing modules on
the data viewer dashboard. Multiple files can be viewed simultaneously and replaced with another
file as needed. Video players are also built into each module so they can be played real-time. Based
on the expertise and needs of biomechanics experts, coaches, and athletes, the dashboard contains
four data viewing module blocks, each displaying a different data multimedia file. In the example
below in Figure 5.3, the user has selected to view data for the same athlete in Figure 5.2 above
(athlete name, date, locations have been de-anonymized for privacy). Using the column on the left
of the data viewer, the user can select the collection to view data for, and in doing so it populates
the list below with all of the available data types and files for trials of that collection. By selecting
one of the 4 colored data viewing module blocks and then selecting the desired data file opens the
file in that square. Video players and image viewers are built-into the data viewer interface. In the
example, shown the user has selected to play the hop and jump event videos for Triple Jump 1 and
below they can simultaneously review the Force Time Plots for the specific hop and jump events.

75

Figure 5.3: This figure shows the data viewing dashboard that allows for advanced simultaneous
data analysis. On the left column, users can select the collection they want, and in doing so it will
populate the list of available data types and files for the trials of that collection. Then by clicking
any one of the 4 colored data viewing module blocks and clicking a data file, the data file will be
opened instantly in the selected block. In this example, it shows the ability to play videos of the hop
and jump of the Triple Jump Trial 1 while simultaneously reviewing the Force Time Plot for the
hop and jump as well.
The layout can also be customized into other templates that were designed together with
the biomechanics research lab for displaying advanced multimedia such as strobe overlay images,
which have elongated rectangular shapes. There are 2 additional possible layouts: two blocks and
two elongated rectangles, or four elongated rectangles. Figure 5.4 below shows the user using a
customized layout to view specialized data files. The top two colored blocks have been customized
into two elongated rectangles which are perfect for viewing elongated film strip or strobe data. In
the figure, the user is viewing the Force Vector Overlay Videos for the Triple Jump Step event for
Trials 1 and 2 while simultaneously reviewing the Force Vector Filmstrips for each of the trials.

76

Figure 5.4: The data viewer dashboard layout can also be customized using premade layout
templates. In this example, the top two colored blocks have been customized to elongated rectangles
which are perfect for viewing elongated Force Vector Film Strips. The user is able to view the Force
Vector Overlay Videos for the Step event of Trials 1 and 2 while simultaneously reviewing the
Force Vector Film Strips for those trials.
The customizable data viewer gives the user the capability to review different forms of data
simultaneously from any collection for comparison and analysis. This data is retrieved real-time
and displayed almost instantly on the viewing dashboard. This gives researchers, coaches, and
athletes from USATF the ability to perform much more sophisticated analysis on a single
streamlined user interface and not only quickly retrieve any data needed, but also analyze the data
in brand new ways.
The web-based platform provides another advantage for the USATF project because while
it creates a streamlined and effective workflow, it still allows for the flexibility required in the
research project processes. When the biomechanics researchers are meeting with coaches and
athletes in person at a collection or competition, they can use the IBIS platform to view the data

77

together. However, even when researchers, coaches, and athletes are not in the same location they
can all remotely access the data through the web-based platform.
5.1.4 - Knowledge Discovery/Decision Support
The Data Review and Analysis steps lead to the ultimate goals of biomechanics research:
knowledge discovery and decision support. For the USATF project, the integration of the IBIS
platform into the biomechanics research workflow provides the many advantages to allow
researchers, coaches, and athletes to quickly identify key findings for improving performance and
preventing injury. As discussed earlier, the IBIS-integrated workflow is much more streamlined
for researchers to easily upload and manage data for coaches and athletes to review all in one
centralized web-based platform.
The customizable data viewer is a novel tool in biomechanics research field for track and
field allowing for simultaneous real-time analysis of multimedia data from the athlete’s history.
By reviewing jump trial videos, force images, force vector overlay videos, and more all together
on one dashboard, researchers, coaches, and athletes can all analyze the data in new ways to gain
deeper insight.
The tool that plays a large role in sharing findings and decision support is the built-in
Annotation/Collaboration tool that allows researchers to leave and save notes on any key findings
they discover from reviewing videos/images or using the statistical analysis tool. Coaches and
athletes are also able to use this tool to save any important findings they discover. For example,
the athlete can record their thoughts and experiences during the trials and the coaches can record
their feedback on the trials as well. Then they can write down and save any questions for further
in-depth analysis on the module for the biomechanics researcher. The module also serves as a hub

78

for collaboration within the USATF project. Because the USATF track and field athletes and the
coaches are often very busy with their training and competition schedules, they may not always be
available to meet in person or remotely with the biomechanics researchers to discuss new findings.
However, the collaboration module solves this issue because with it, the biomechanics researcher
can save any notes on discoveries, videos, data to review on the module and when the coach/athlete
is free they can log onto the IBIS platform themselves and review those notes. After reviewing,
they can make any changes and adjustments to the training regimen or technique, effectively
leading knowledge discovery to decision support. Any additional feedback or responses from the
coaches and athletes is then also recorded on the collaboration module, providing a tool for
efficient back and forth communication between all project members in one central location with
the data.
One of the most powerful tools on the IBIS is the built-in statistical analysis dashboard,
with which all users can perform high level analysis on the collected data without the need of
downloading and installing a third-party software. Biomechanics researchers can use the
streamlined user dashboard with custom filters and selections to explore key findings and share
with USATF coaches and athletes. By using the filters, selections, and identifiers, the users can
explore countless combinations of data to visualize on the graphical interface shown in Figure 5.5
below. Data points plotted on the graphical interface are based on the user’s variable and identifier
selections. In the example, the user is visualizing changes in velocity for the hop, jump, and step
stages of triple jump trials for a set of athletes (shown by the color identifiers). The changes in
velocity are plotted over time for longitudinal analysis. By tracking these variables, it provides
detailed insight into each of the jump trials which can be linked to which of the jumps had better
results or resulted in pain/injury. The effects of having higher or lower velocities at specific parts

79

of the triple jump varies depending on training expertise from the coach, anatomy of the athlete,
and other factors. Using Athlete 2 as an example, (shown by the green dots) in the Jump phase,
there are years where there are larger changes in velocity. Depending on coaching or training
methods, this would be the areas that the coaches and athletes would like to explore in further
detail. By selecting the data points, the user can open the associated multimedia for those trials
with the data viewers shown in Figure 5.2. This knowledge discovery leads to further discussion
and analysis with the biomechanics lab on other associated data from those same trials. This
information in addition to the performance results such as jump distance allows the coaches to
make adjustments as needed to improve results or prevent injury. This decision support allows
coaches and clinicians to work with athletes in an approach similar to personalized medicine, by
fine-tuning training regimens, technique, and more. Because the statistical analysis tool is
integrated with the IBIS system and web-based, it can be accessed from the biomechanics lab or
remotely at collection/competition sites with internet access. This is an immense functional
improvement compared to manually opening and searching through multiple log sheets to acquire
this data. The instant data retrieval and visualization allow for real-time complex analysis of track
and field jump data along with all available videos/images to provide insight and new knowledge
for coaches and athlete to make adjustments right away and improve performance or prevent
injury.

80

Figure 5.5: This displays the statistical analysis tool integrated within the IBIS system. By using
filters, selections, and identifiers, the user is visualizing the changes in velocity for the hop, jump,
and step portions of the triple jump. For each athlete (identified by color), they are able to track their
measurements over time. This information provides insight into which jumps resulted in better
performance and further analysis of those jumps, leading to adjustments in training and technique
from the coach. For example, when looking at Athlete 2, shown by the green dots, during the Jump
phase (middle row of data), it can be seen which years contained larger changes in velocity.
Depending on coaching or training methods, this would be the areas where the coaches, athletes,
and researchers would want to dive deeper into analysis by pulling up the multimedia data viewers
previously shown in Figure 5.2
5.2 - PAC12 Long Distance Running Stress Fracture Study
For the second use case, the IBIS was integrated into the PAC12 project research workflow
with an emphasis as a platform for streamline multi-institutional collaboration. Because the
multiple participating institutions are located in different locations and used different
software/methods, there were challenges in the integration and analysis of data. This displayed the
ability of IBIS to support individual as well as collaborative projects with multiple institutions.
The new research workflow integrated with the system and support tools was tested with the data
collected from and at all 4 collaborating research labs. All key steps in the workflow are discussed

81

in detail in each of the following sections with workflow comparisons as shown in Figure 5.6
below.

Figure 5.6: A comparison between the original PAC 12 research workflow and the new workflow
with the IBIS system. In the original workflow, each biomechanics lab has their own locations for
storing data. These labs then migrate the data over to Google Drive and email to share the data with
coaches, athletes, and researchers from collaborating institutions. Each institution has to set correct
user access privileges for shared folder/files on Google Drive and track/manage email threads. The
IBIS integrated system significantly improves the workflow by streamlining it through one
centralized platform where all tasks/steps can be accomplished.
5.2.1 - Data Collection/Preprocessing
The 4 collaborating research groups at USC, Colorado, Oregon, Stanford work closely with
the track and field team at their school to collect data for the long distance running study. The
study data for USC was collected on an outdoor track site and the data for the 3 other schools were
collected on instrumented treadmills. Similar to the USATF project, collected data can be
immediately uploaded and stored into IBIS system database or done so after pre-processing as
well. The data can be retrieved whenever needed for cleaning and pre-processing which is also
essential in the PAC12 study in order to standardize the data for collaboration. The centralized

82

IBIS platform and database promotes and streamlines the standardization and storage of the data
from all institutions such as the cropped the treadmill data representing each step to be integrated
with the data of each step from the overground force plate data. The preprocessing tools such as
the automatic foot detection tool is continuously being developed and tested on the data for key
event identification. These tools greatly impact the future of time-syncing data as well as
applications for cropping treadmill data to streamline and enhance the data pre-processing steps.
5.2.2 - Data Upload/Storage
The web-based IBIS platform for the PAC 12 project is also installed on the server located
in the USC EGG Data Center to ensure secure connections monitored by USC ITS. The ITS
firewall protects the web security of the IBIS system as well as monitors for any potential new
cyberattacks of suspicious activity. Users from all institutions were supplied user credentials to
remotely log-in to the system with internet access. Only authenticated users are allowed to access
the platform and upload/download data. The system includes user access controls set by the system
administrator which assigns specific access rights to each user controlling exactly what data the
user can view and also what actions the user can perform on the data. For example, USC coaches
and researchers will be able to see all the data for only USC athletes, and similar for other schools.
In addition to a location for shared data, there are also locations for each school to privately
warehouse their desired data.
The PAC12 data uploaded by the 4 research groups is all located in one centralized location
for streamlined retrieval and sharing. The different groups no longer need to use different storage
solutions for saving or sharing their data. In the IBIS system there is a standardized organization
of the data based on the ePR framework. Data is organized by athlete so that researchers, coaches,

83

and athletes can quickly located and select the desired athlete subject. All of the collections and
data for every trial for that athlete will be readily available. The standardized raw and calculated
data values and discrete variables are uploaded and stored in the MySQL database of the IBIS
system. The visualization and analysis of these variables will be discussed further in the following
sections.
5.2.3 - Data Review/Analysis
Once the data has been uploaded into the centralized IBIS platform, all authenticated users
can remotely access the server to download, review, analyze the data. Data can be visualized using
the built-in multimedia tools such as the image viewer and video player. Once an athlete subject
is created on the IBIS system, researchers from all institutions can upload data for sharing using
the upload modules built into the web-GUI. As shown in Figure 5.7 below, multiple modules can
be created for uploading shared data and these modules are robust and flexible enough to perform
a variety of functions. They can be named as needed and function locations for uploading similar
data types or subsets of data addressing a specific research question.

84

Figure 5.7: This is the user interface dashboard of a shared location for a de-anonymized athlete
where collaborating researchers from all institutions can upload and share data with each other. The
different modules can be used to upload groups of similar data types or subsets of data addressing
certain research questions.
The user interface dashboard for de-anonymized shared athletes and for private
warehousing by each institution are identical in functionality. By clicking view the user can
instantly open a multimedia file such as a video or an image for review without the need to
download or open third-party software. The flexibility of the IBIS platform is essential in providing
the ability to securely upload while protecting the privacy of that data and also the ability to
accurately control and allow quick efficient data sharing and collaboration. This ability allows for
a novel streamlined collaboration process that is otherwise not possible without the IBIS system.
Similar to the data viewer in the USATF project, the data viewing module is also available
in the PAC 12 project. The built-in data viewing module allows users to select an athlete and then

85

view all the multimedia data simultaneously on a single streamlined user interface. Files such as
recorded video and force curve images can be instantly pulled on each of the data viewing module
blocks within the viewer. The template layout of the data viewer can also be adjusted to best match
the needs of the data being displayed. In order to review and analyze multimedia, all of the
collaborating research groups can perform the necessary actions on the centralized IBIS platform,
which houses all of the shared data. The times to collect data at each institution and discuss findings
vary by time zone as well as training session/collection/competition schedules. The web-based
remote capabilities of the IBIS platform have a huge impact with its time efficiency and flexibility
to match the needs of the biomechanics study. Researchers can upload data during collections if
time allows or after, review current and old data with coaches and athletes right away, and
collaborate with other institutions regardless of location on any computer or laptop device
whenever there is internet connectivity.
5.2.4 - Knowledge Discovery/Decision Support
As mentioned earlier, in the PAC12 project there is an added emphasis on streamlined
multi-institutional collaboration and data exploration. The ultimate goal of the project is to
discover new knowledge about the development of stress fractures in long distance running
through data sharing and analysis. As discussed in the previous sections, the IBIS platform has
already been able to provide a secure and centralized location for users from all institutions to
share and review multimedia data together.
The integrated statistical analysis tool in the IBIS system plays a huge role in the PAC 12
study for visualization of raw and calculated variables stored in the IBIS database. The tool is
built-in and web-based so researchers from any institution can access the tool remotely without

86

the need to download and install any new software. This solves the issues of researchers from
different institutions using different programs to perform analysis and compatibility issues when
sending data to another researcher who uses another program. The R Shiny application for
statistical analysis contains a streamlined user dashboard with different filters and buttons for data
query as shown in Figure 5.8.

Figure 5.8: This is the user interface dashboard for the integrated R Shiny application on the IBIS
system for statistical analysis. The example is using the statistical tool for knowledge discovery in
the PAC12 project. By using the selectors on the left column of the dashboard, the x-axis is
representing vertical force and the y-axis is representing contact time. The analysis shows that as
pace as increases (faster pace represented by red and green markers), vertical force increases and
contact time decreases based on the x and y-axis. This discovery shows that when running at
different paces with the same number of foot strikes there will be higher average loads which may
cause more stress fractures.
In the PAC12 project example shown above, the user has set the x-variable, y-variable, and
color code variable to analyze the relationship and correlations between pace, vertical force, and
contact time in long-distance running steps. After selection, the data is queried and visualized on
the GUI in real-time. This functionality allows the PAC12 team to quickly analyze any
combination of data sets to explore research questions and discover the right questions to ask. In
the example, the pace is shown by the color identifier on the right side. Faster paces are represented

87

by the green and red data point markers and for these points there are higher vertical forces (x-
axis) and lower contact times (y-axis). For faster paces it is shown that vertical forces will be
greater and contact times will be shorter. In comparing two runs at different paces of 100 total foot
strikes, this leads to the conclusion of higher average loads for the shorter run. With this discovery,
the biomechanics team is able to create a new question to explore: whether or not higher average
loads lead to higher occurrence of stress fractures. This allows the research team to provide
feedback to coaches and athletes to see if adjustments need to be made and for continued data
collection/analysis.
Another example of the R Shiny statistical tool in use in the PAC12 project is shown below
in Figure 5.9. The researchers are using the dashboard to perform cross-study analysis by
visualizing the relationship between pace and contact time across multiple study subjects. The data
can also be filtered by other factors such as weight, date, and injury range shown with the filters
below the plot. The different subjects are displayed on the x-axis. The purple and blue markers
representing slower pace are associated with longer contact times (larger values on the y-axis).
However, it can be seen that there is variation between different subjects and also within a subject
themselves. Longer contact times may be caused by differences in foot strength and ability to
regulate movement and the subjects having overall longer contact times can be quickly identified
as having their spread of data points located higher on the y-axis. The visualization allows the
biomechanics team pick up on individual differences and go back to look at additional details for
these subjects such as ground reaction forces, whether they are a chronic heel striker or fore foot
striker, and see if these are the causes for stress fractures. By selecting the data points, the user is
able to open the associated multimedia data for those trials with the built-in data viewers instantly.

88

Figure 5.9: This is another example of the built-in IBIS statistical analysis tool in use for the PAC12
research study. Users use the dashboard to instantly query data across multiple participant subjects
for cross-study analysis. The example shows that although longer contact times (y-axis) are
associated with slower paces (purple/blue color identifiers), there is variability between subjects (x-
axis) and also within the same subject. Longer contact times may be caused by differences in foot
strength/regulation and with the tool, the subjects with overall longer contact times (spread of data
points is higher on the y-axis) can be quickly identified for further analysis by studying their ground
reaction forces, type of foot strike, etc. These data points can be selected to open the associated
multimedia data.
The data viewer and R Shiny dashboards allow PAC12 users to quickly retrieve both
multimedia and discrete data in real-time for analysis and knowledge discovery. With these tools
connected to the IBIS database, research teams including athletes, coaches, researchers, and
clinicians can find the right questions to explore on a centralized platform without the need to
install or learn how to program with different software programs.
In the first example in Figure 5.8, through the data exploration in the R Shiny statistical
analysis tool, it was discovered that as pace in long distance running increased, there were
increases in vertical forces and decreases in contact time. This leads to the finding that there are
higher average loads which may lead to stress fractures. The second example in Figure 5.9 shows
through cross-study analysis for multiple subjects the variability in the relationship between pace

89

and contact time between individuals and within individuals. The tool allows the quick
identification of subjects with longer contact times to further study individual differences such as
foot strength, ground reaction forces, etc. The R Shiny app directly leads to findings to be shared
and tested with coaches and athletes while promoting the discovery of novel areas of exploration.
These new areas to explore are brought to light by the analysis tool and the continued analysis will
be performed with the same tool streamlining and brining the research workflow full circle.
The IBIS platform for the PAC12 also provides the same communication module used in
the USATF project for streamlined collaboration. Since all multimedia/raw/processed data, data
viewers, tools are located on a centralized IBIS system, research teams at all locations can
review/analyze the data and while recording their thoughts and findings into the communication
module. This module functions as a central hub for multi-institutional collaboration as findings
can be saved and shared for with all authorized users efficiently. Because the system is web-based,
the module is readily available for researchers from all institutions to access at all times from lab
locations, training facilities, competition sites, or any location with internet access. The 4
participating schools no longer have to keep track of research discussions, questions, or findings
using different methods of communication.
5.3 - Workflow Efficiency Comparison
In order to grasp the advantages and impact of the IBIS system, it is essential to look at the
workflow timing and efficiency with and without the integrated informatics system and its tools.
Sets of tasks that are essential in performing biomechanics research studies in track and field were
designed with the USC Biomechanics Lab. These tasks are shown in Table 5.1 below and include
the estimated times to complete the tasks with and without the IBIS system. The estimated times

90

include a range of time due to the varying backgrounds and expertise of the users, which include
biomechanics experts, research interns, coaches, and athletes.
Workflow Task Estimated
Time
Estimated
Time with IBIS
Locate and open a video file for a specific trial and
collection
5 min – 2 hr < 10 sec
Locate and open video and force files for a specific trial
and collection
10 min – 4 hr < 10 sec
Perform pre-processing step of finding frame of foot
contact for all trials in a collection
2 hr – 5 hr < 30 min – 1 hr
Plot and display any single variable over time 10 min – 1 day < 10 sec
Plot and display a set of variables for multiple subjects 15 min – 1 day < 10 sec

Table 5.1: This table shows the workflow tasks regularly performed within the biomechanics
research process for track and field projects. Each task includes the estimated time of completion
with and without the IBIS system and its tools.
The first two workflow tasks involve locating a video file and a force file for a specific
athlete’s trial from a certain collection. For a very experienced researcher, the estimated amount
of time can range from a few minutes to a few hours because the data can be stored on Google
Drive or physical drives that the user needs to manually look through and search for the appropriate
data files. For a researcher who is newer to the biomechanics lab or a coach/athlete who does not
work in the lab, the estimated time to find these files may be much higher or never completed if
the files cannot be located. With IBIS system, the users, regardless of experience, only need to log
into the system, select the athlete, collection, trial, and files to view with a few clicks, lowering the
estimated task time to a few seconds.
The third workflow task involves the pre-processing of video data involving foot contact
detection. In order to go through every trial in a collection and manually go through all the frames
to select the frame of foot contact takes a few hours. With the IBIS system and the automatic foot
contact detection tool prototype, the user only has to look through 5 frames to select the frame

91

where foot contact occurs for each trial, significantly lowering the estimated time needed to an
hour or less.
The fourth and fifth tasks involve the locating variables recorded a subject over all time.
This data can be located in different locations such as Google Drive and physical hard drives or
USB sticks. The data may also be collected over many years and so the folder organizations and
naming conventions for the folders/files may be completely different. Then once all the data is
located, it needs to be manually imported into a software to run or program the code to plot and
display the data. For an experienced researcher, the time needed to complete this task may range
all the way to an entire day. As for the level of programming or software knowledge necessary for
this task, it may take much longer or unable to be completed by research interns, coaches, or
athletes. With the IBIS system and the integrated statistical analysis tool, there is no programming
or software knowledge needed. All of the data will be stored in the IBIS database as well, so all a
user needs to do is log in, select the athlete(s), and variables to visualize in real-time, which takes
less than a few seconds.
Through the workflow efficiency comparisons of these essential tasks in the biomechanics
workflow for track and field research, the times needed for the tasks are significantly lower with
the integrated IBIS system. As discussed earlier, there are also situations in which certain tasks
can’t be completed without the IBIS system.

92

Chapter 6: Discussion and Future Work
In the previous chapters I have discussed the background of biomechanics and the sports
data research workflow specifically in track and field sports. Upon identifying the limitations, a
multi-institutional integrated biomechanics informatics system was designed based on proven core
principles in medical imaging informatics. These principles protect and streamline the
biomechanics research workflow ensuring accessibility, efficiency, security, and privacy, while
introducing novel data analysis and statistical built-in tools. The IBIS system and workflow were
integrated into two use case projects: the USATF Long Jump and Triple Jump Study and the PAC
12 Track and Field Study On Stress Fractures In Long Distance Running. With the IBIS system,
there were immense improvements in functionality and efficiency, as well as providing novel
methods of data review, analysis, sharing, and collaboration.
The introduction of a centralized IBIS system for data management greatly improved and
streamlined both the USATF and PAC12 research workflows. After collections are completed,
data can be securely uploaded right away into the IBIS server to act as a data warehouse. Not only
is all the data centralized in a single location, the location can be accessed remotely regardless of
location as long as there is internet access. The IBIS upload and storage provides the flexibility
needed in biomechanics research studies. The data can be uploaded at the collection site, or if there
is no internet access at the site, later at another location or back in the laboratory. The flexibility
also had advantages managing data at different stages. For example, raw data can be downloaded
for pre-processing and uploaded back into the server to retrieve later for post-processing. Support
tools such as the automatic foot detection were used in pre-processing steps for the use cases saving
large amounts of time spent on manually processing the data. As shown in the Section 5.3, the key
workflow step of locating frame of foot contact for every trial in a collection originally took up to

93

5 hours. With the automatic foot detection tool integrated into the workflow, the time needed for
the task is reduced to 30 min to an hour with the same level of accuracy. This IBIS system
introduces a unique combination of robust features but at the same time allows for flexibility
demanded in the biomechanics research field depending on each research lab’s practices,
workflow, site conditions, and environmental conditions.
The IBIS platform includes a built-in streamlined GUI user interface for uploading,
managing, and downloading data from the centralized server database. As shown in both use cases,
users were able to open video and image files for review using built-in functionalities without the
need to download, install, or open any additional software. The robust design supports multi-
institutional collaboration with privacy and user access controls that are set by the system
administrator. User access privileges control which authenticated users are allowed to view and
manage a specific set of data. As shown in the PAC12 use case, specific data sets on the system
are privately warehoused for each institution, while certain data sets are shared specifically for
collaboration. The IBIS system eliminates the need to share physical hard drives or to download
and install any third-party software in order to upload/manage/share data. Because the system is
accessible remotely from a centralized web-based application, there is no need for different users
to share data and assign specific access rights to each file/folder in their respective programs. This
can all be set up in one location by the system administrator. This single centralized platform
greatly increases the efficiency of IBIS-integrated research workflows in biomechanics by
drastically reducing the time required to perform key workflow tasks in data analysis as well. Tasks
such as locating and opening files, which usually range from 10 min to hours to complete, are
reduced to times of less than 10 seconds. There are situations where physical drives containing the
necessary data can’t be located and in this case the task can’t even be timed. This situation is

94

avoided when all the data is stored in the IBIS database. When factoring the multitude of these
tasks performed in research every day, the time saved is immense.
The streamlined data viewer dashboard was used in both use cases to open multimedia files
from different collections almost instantaneously for simultaneous viewing. A variety of
multimedia including videos and images were queried and displayed in real-time for review and
analysis. Customized data viewing template layouts allowed for users to effectively display unique
data types such as the elongated strobe overlay images in the USATF project for long jump/triple
jump. This data viewer dashboard gives all users the capability to instantaneously review data
simultaneously in a streamlined fashion that is otherwise not possible promoting knowledge
discovery and decision support. Viewing this data in new ways allows teams of researchers,
athlete, and coaches to ask the right questions, find solutions, and adjust training/techniques to
prevent injury and improve performance.
The R Shiny statistical analysis tool provided a novel web-based GUI for users to analyze
the collected and processed data without the need to download/install/learn additional stats
analysis software. Users are able to select and compare different sets of variables without any prior
programming experience. The data is stored in and linked the IBIS database so that they can be
queried real-time for quick analysis at the research lab, collection site, or even competition
location. In the PAC12 use case, this functionality was invaluable in solving the issue of different
research institutions performing statistical analysis with a variety of software programs. Instead of
expending extra time for each group’s researchers, athletes, coaches to learn how to use each
program, all users were able to use the built-in statistical analysis tool with streamlined dashboard
filters and toggles. The interface allows all collaborating groups to explore data longitudinally over
time as well as perform cross-study analysis to search for correlations across multiple subjects. In

95

the USATF example, the user was able to visualize the changes in velocity for each of the stages
of the triple jump trials longitudinally. This information provided more detailed insight into the
performance for the athletes for each of the jumps and allows the team to determine which jumps
of interest to further explore. This leads to more personalized adjustments in the training regimen
and technique. In the PAC12 example, the user used the built-in statistical analysis to discover that
for long distance running, as pace increased, vertical force increased and contact time decreased.
The tool was able to show this for multiple athletes at the same time for cross-study analysis. These
findings show that for running at different paces, but with the same number of foot strikes, there
is a higher average load, which could lead to higher incidence of stress fractures. The novel
information is discovered with the efficient and software-less R Shiny tool and allows the research
team and clinicians to provide decision support for athletes through a streamlined workflow.
These tasks of visualizing data in the research workflow used to take anywhere from 15
min for an experienced programmer to an entire day or more for a user with no programming
experience. With the web-based built-in application, the tasks can be completed in less than 10
seconds by users with any level of or no programming experience at all drastically improving
functionality and efficiency. Users have the ability to quickly explore the same data in real-time,
querying specific subsets of data and combinations or variables on the same platform. Not only is
data sharing and analysis performed in one centralized location, but also collaboration and
discussion as any findings can be documented using the collaboration module also built into the
IBIS web platform. There all users can share any findings or pose any research questions for the
entire group to examine. As projects expand, the IBIS system can easily add additional institutions
and their data. Having a single location that can be remotely accessed to perform all the tasks

96

necessary in a collaborative research effort is an indispensable tool for knowledge discovery and
decision support.
The future potential for the IBIS is endless as it can revolutionize informatics for
biomechanics and sports in the same manner that PACS did for Radiology and the Healthcare
Enterprise. The potential roadmap for the future includes the expansion to all other sports where
streamlined data integration and a multi-disciplinary effort leads to knowledge discovery in both
injury prevention and performance improvement. Future work also includes expanding the
platform to support studies for clinical rehabilitation subjects such as for wheelchair users.
Measurements including force/acceleration data as well as associated video data can be integrated
into the system for streamlined data management and analysis.
The ultimate goal is to integrate IBIS with the outcome data. Once researchers and
clinicians have worked with coaches and athletes to discover key findings for preventing injury or
improving performance from the IBIS, the next step will be to make adjustments based on these
findings. This decision support may lead changes in training regimen, warmup routines, technique,
etc. The goal is to then collect the data from additional trials after making the changes and analyze
the data for findings that support outcomes and/or further adjustments, starting the cycle again.
With the integration of outcomes, everything will come full circle. The ability to centralize and
support all the needs of multiple research iterations shows the IBIS is a platform of continuous
discovery and improvement.
The novel IBIS platform and its tools lead to newfound discoveries on the biomechanical
data which are communicated to the coaches and athletes. With this data, coaches and athletes
make informed decisions based on the biomechanical findings to improve performance and
prevent injury. The translation from knowledge discovery to decision support has huge potential

97

impacts beyond track and field for biomechanics and other sports including basketball, baseball,
and more. Through the development and evaluation, the IBIS system has become a platform that
supports multi-institutional collaboration and acts as a centralized hub for data storage,
management, review, and analysis that can be securely access remotely at any time.

98

References

1. Sykora, M., Chung, P., Folland, J., Halkon, B., and Edirisinghe, E, “Advances in sports
informatics research,” Computational Intelligence in Information Systems. Volume 331, 265-
274 (2015).

2. Liebermann, D., Katz, L., Hughes, M., Bartlett, R., McClements, J., and Franks, I., “Advances
in the application of information technology to sports performance,” Journal of Sports
Sciences. Volume 20(10), 755-769 (2002).

3. Link, D., and Martin L, "Sport informatics-historical roots, interdisciplinarity and future
developments," International Journal of Computer Science in Sport. Papers 8(2), 68-87
(2009).

4. Mcnitt-Gray, J., Director, USC Biomechanics Research Lab, meeting, Jan. 2018

5. O’Donoghue, P., “Research Methods for Sports Performance Analysis,” Routledge, 2009.

6. Huang, H. K., Pacs and Imaging Informatics: Basic Principles and Applications. 2
nd
ed. New
Jersey: John Wiley & Sons, Inc.

7. Wang, X., Liu, B., Martinez, C., Zhang, X., and Winstein, C., “Development of a novel
imaging informatics-based system with an intelligent workflow engine (IWEIS) to support
imaging-based clinical trials,” Computers in Biology and Medicine. Volume 69, 261-269
(2016).

8. Verma, S., McNitt-Gray, J., and Liu, B., “Medical imaging informatics based solutions for
human performance analytics”, Proc. SPIE 10579, Medical Imaging 2018: Imaging
Informatics for Healthcare, Research, and Applications, 105791R (6 March 2018).

9. Pallis, J., McNitt-Gray, J., Hung, G., Biomechanical Principles and Applications in Sports. 1
st

ed. Switzerland: Springer Nature Switzerland AG.

10. Vincent, J., Pro Stergiou, Katz, L., “The Role of Databases in Sport Science: Current Practice
and Future Potential.” International Journal of Computer Science in Sport. 8(2): 50–66, (2009).

11. Le Meur, Y., Torres-Ronda, L., “10 Challenges Facing Today’s Applied Sport Scientist.”
Science Performance and Science Reports. 57: 1–7, (2019).

12. Deshpande, R., DeMarco, J., Sayre, J., Liu, B., “Knowledge-driven decision support for
accessing dose distributions in radiation therapy of head and neck cancer,” International
Journal of Computer Assisted Radiology and Surgery. Volume 11, 2071-2083 (2016).

99

13. Ma, K., Liu, B., Zhou, Q., McNitt-Gray, J., “Imaging informatics-based electronic patient
record and analysis system for multiple sclerosis research, treatment, and disease tracking,”
University of Southern California Library (2018).

14. Lee, S., Hidler, J., “Biomechanics of overground vs. treadmill walking in healthy
individuals,” Journal of Applied Physiology. Volume 104(3), 747-755 (2008).

15. Zeni Jr., J., Richards, J., Higginson, J., “Two simple methods for determining gait events
during treadmill and overground walking using kinematic data,” Gait & Posture. Volume
27(4), 710-714 (2008).

16. Winter, D., Biomechanics and Motor Control of Human Movement. 4
th
ed. New Jersey: John
Wiley & Sons, Inc (2009).

17. Asmussen, M., Kaltenbach, C., Hashlamoun, K., Shen, H., Federico, S., Nigg, B., “Force
measurements during running on different instrumented treadmills,” Journal of
Biomechanics. Volume 84, 263-268 (2019).

18. Roberts, B., Hall, J., Williams, A., Rouhani, H., Vette, A., “A method to estimate inertial
properties and force plate inertial components for instrumented platforms,” Medical
Engineering and Physics. Volume 66, 96-101 (2019).

19. Hnat, S., van Basten, B., van den Bogert, A., “Compensation for inertial and gravity effects
in a moving force platform,” Journal of Biomechanics. Volume 75, 96-101 (2018).

20. Nelson, R., Dillman, C., Lagasse, P., Bickett, P., “Biomechanics of overground versus
treadmill running,” Medicine and Science in Sports and Exercise. Volume 4(4), 233-240
(1972).

100

Appendix

DICOM – Digital Imaging and Communications in Medicine
eAR – Electronic Athlete Record
ePR – Electronic Patient Record
GUI – Graphical User Interface
HIPAA – Health Insurance Portability and Accountability Act
HIS – Hospital Information System
HL7 – Health Level 7
HTML – Hyper Text Markup Language
IBIS – Integrated Biomechanics Informatics System
IHE – Integrating the Healthcare Enterprise
IWEIS – Intelligent Workflow Engine System
LAMP - Linux, Apache, MySQL, PHP/Perl/Python
MIII – Medical Imaging Informatics Infrastructure
MySQL – “My’ Structured Query Language
PAC12 – Pacific Coast Conference
PACS – Picture Archiving and Communication Systems
PHP – Hypertext Preprocessor
RIS – Radiology Information System
TCP/IP – Transmission Control Protocol/Internet Protocol
USATF – United States Track and Field
WADO – Web Access to Dicom Objects

Abstract (if available)

Abstract The field of biomechanics involves integrating a variety of data types such as waveform, video, discrete, and performance. These different sources of data must be efficiently and accurately associated to provide meaningful feedback to athletes, coaches, and healthcare professionals to prevent injury and improve rehabilitation/performance. There are many challenges in biomechanics research such as data storage, standardization, review, sharing, and accessibility. Data is stored in different formats, structures, and locations such as physical hard drives or Dropbox/Google Drive, leading to issues during sharing and collaboration. Data is reviewed and analyzed through different software applications that need to be downloaded and installed locally before learning how to use. An integrated biomechanics informatics system (IBIS) built based on the core principles in medical imaging informatics provides a solution to these many challenges. The system provides a secure web-based platform that will be accessible remotely for authenticated users to upload, share, and download data. The web-based application includes built-in data viewers that are streamlined for reviewing multimedia data and decision support/knowledge discovery tools. These tools include automatic foot contact detection for pre-processing, built-in statistical analysis applications for longitudinal and cross-study analysis, and a multi-institutional collaboration module. The IBIS system creates a centralized hub to support multi-institutional collaborative biomechanics research and analysis that is remotely accessible to all users including athletes, coaches, researchers, and clinicians generating a novel streamlined research workflow, data analysis, and knowledge discovery process.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A medical imaging informatics based human performance analytics system

PDF

Identifying injury risk, improving performance, and facilitating learning using an integrated biomechanics informatics system (IBIS)

PDF

Imaging informatics-based electronic patient record and analysis system for multiple sclerosis research, treatment, and disease tracking

PDF

Knowledge‐driven decision support for assessing radiation therapy dose constraints

PDF

Decision support system in radiation therapy treatment planning

PDF

Molecular imaging data grid (MIDG) for multi-site small animal imaging research based on OGSA and IHE XDS-i

PDF

Mining an ePR system using a treatment plan navigator for radiation toxicity to evaluate proton therapy treatment protocol for prostate cancer

PDF

Integration of digital twin and generative models in model-based systems upgrade methodology

PDF

Understanding reactive balance control strategies in non-disabled and post-stroke gait

PDF

Using biomechanics to understand how an individual accomplishes a task in a variety of contexts

PDF

2D ultrasonic transducer array’s design and fabrication with 3D printed interposer and applications

PDF

Advanced applications of high frequency single beam acoustic tweezers in fluid and cell mechanics

PDF

Architectural innovations for mitigating data movement cost on graphics processing units and storage systems

PDF

Data-efficient image and vision-and-language synthesis and classification

PDF

Pediatric magnetic resonance image processing: applications to posterior fossa cancer and normal development

PDF

Decoding memory from spatio-temporal patterns of neuronal spikes

PDF

Knowledge, motivation, and organizational influences within leadership development: a study of a business unit in a prominent technology company

PDF

Developing an agent-based simulation model to evaluate competition in private health care markets with an assessment of accountable care organizations

PDF

Pathomechanics of femoroacetabular impingement syndrome: utilizing subject-specific modeling approaches to investigate the influence of hip joint morphology and neuromuscular control

PDF

Toward better understanding and improving user-developer communications on mobile app stores

Asset Metadata

Creator Liu, Joseph William (author)

Core Title Development of an integrated biomechanics informatics system (IBIS) with knowledge discovery and decision support tools based on imaging informatics methodology

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Degree Conferral Date 2022-08

Publication Date 08/03/2022

Defense Date 04/20/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag biomechanics,Biomedical Engineering,biomedical imaging,data exploration,data integration,data management,data visualization,health information systems,imaging,informatics,informatics systems,knowledge discovery,multimedia analysis,OAI-PMH Harvest,research workflow,sports data,sports data analysis,sports medicine,sports video,statistical analysis,user interface

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Liu, Brent (committee chair), McNitt-Gray, Jill (committee member), Zhou, Qifa (committee member)

Creator Email joseph9wl@gmail.com,josephwl@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111375989

Unique identifier UC111375989

Legacy Identifier etd-LiuJosephW-11079

Document Type Thesis

Format application/pdf (imt)

Rights Liu, Joseph William

Type texts

Source 20220803-usctheses-batch-967 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu