Adapted PE and Computer Science Team Awarded ORED Multidisciplinary Grant
It can be a challenging, complex process placing a child with an injury or disability into a physical therapy or adapted physical education service. That is a problem Dr. Melissa Bittner, Associate Professor in the Department of Kinesiology, knows all too well.
“Typically, part of that process involves a child having to undergo a series of movement tests, in order to demonstrate need for that service,” Dr. Bittner said.
The Test of Gross Motor Development, Third Edition (TGMD-3)1 is the most widely used standardized instrument for assessing fundamental motor skills in children between the ages of 3 and 10 (Bittner et al., 2021).2 Trained professionals score the assessments; however, according to Dr. Bittner, the process can be both labor-intensive and time consuming.
Enter Dr. Jelena Trajkovic, Associate Professor in the Department of Computer Science and Engineering, who is teaming up with Dr. Bittner to try and create a new gold standard in the way movement assessments are scored, using an AI-based scoring system.
“It’s not only a great research topic, but also an opportunity to work on the more applied side of machine learning,” Dr. Trajkovic said.
The collaboration between Drs. Bittner and Trajkovic is an exciting new interdisciplinary venture within the College of Health and Human Services and College of Engineering through the ORED Multidisciplinary Grant. The Office of Research and Economic Development (ORED) at is an internal funding opportunity designed to foster collaboration between faculty from different disciplines, typically providing funds to support projects that lead to major external funding, such as federal grants.
“It has been a great fit working with Dr. Trajkovic,” said Dr. Bittner. “We wanted to put together an NIH proposal down the road, and after talking, of course we knew we needed some pilot data, which led us to apply for the multidisciplinary grant [through ORED].”
The project will develop an artificial intelligence (AI)-driven video analysis system, capable of automating TGMD-3 scoring from standard smartphone recordings. Dr. Trajkovic and her student team will begin working to develop a series of biomarkers to “teach the machine.” In other words, this will involve training a “deep learning model” to predict TGMD-3 scores, based on extracted kinematic features. The intended outcome for this computer vision model will be to detect and track key joint movements relevant to four TGMD-3 skills– dribble, strike, horizontal jump, and skip.
Leveraging advances in computer vision, pose estimation, and AI/deep learning, the system will aim to replicate human expert scoring. The hope is to improve objectivity, reliability, and accessibility, while reducing the time and training burden associated with manual scoring. By integrating biomechanics with AI technology, the project contributes to the democratization of motor skill assessment, supporting equitable access for schools, clinicians, and researchers in resource-limited settings.
“This will help us be able to place more kids into services they need and qualify for,” Dr. Bittner said.
Dr. Trajkovic is excited to apply her research expertise and involve students from both colleges to come together to solve a real-world issue.
“I think helping our students gain exposure to real world problems and seeing the practical application to help others is just great,” she said.
“We should talk about bringing your computer science students one day to observe our adapted physical education practicums in action,” Dr. Bittner remarked over a Zoom call with Dr. Trajkovic. “This would allow your students to see the kids with disabilities, and see the ‘why’ of how you are going to help these kids.”
Together, with the assistance of colleague Dr. Ali Brian, Associate Dean for Research and Professor at the University of South Carolina, Drs. Bittner and Trajkovic plan on submitting a competitive NIH R15 grant proposal that integrates pilot data, methodological framework, and collaborative undergraduate student training outcomes.
“I had participated in a Camp Abilities with Dr. Brian in Iceland, and that’s when I pitched my idea to her about this project,” Dr. Bittner said. “From there, I contacted Dr. Shadnaz Asgari, Department Chair of Computer Engineering and Computer Science, and shortly after is when I got connected with Dr. Trajkovic for this multidisciplinary grant.”
“I was really looking into deepening that area of work and research [in application of machine learning and movement analysis], so for me this was super lucky,” Dr. Trajkovic said. “Like a shooting star.”
-
Ulrich, D. A. (2019). Test of Gross Motor Development–Third Edition (TGMD-3). PRO-ED.
- Bittner, M., Foster, E., & Lavay, B. (2021). Assessment Practices in Adapted Physical
Education+. Palaestra, 35(2), 49-54.
