Restricted Research - Award List, Note/Discussion Page
Fiscal Year: 2023
111 University of North Texas (141999)
Principal Investigator: Albert,Mark Vincent
Total Amount of Contract, Award, or Gift (Annual before 2011): $ 289,770
Exceeds $250,000 (Is it flagged?): Yes
Start and End Dates: 1/1/23 - 12/31/25
Restricted Research: YES
Academic Discipline: Computer Science & Engineering
Department, Center, School, or Institute: College of Engineering
Title of Contract, Award, or Gift: Improving Pediatric Orthopaedic Outcomes with Machine Learning
Name of Granting or Contracting Agency/Entity:
Shriners Hospitals for Children
CFDA:
Program Title: none
Note:
Dr. Albert and his two graduate students will coordinate with members of the Motional Analysis Center at Shriners Hospitals for Children – Chicago (Chicago MAC) to further develop the SHC Gait Quality Index score, use it for post-operative outcomes prediction, and apply this approach to address prediction of rehabilitation outcomes and identify relevant features affecting outcomes. Dr. Albert’s team will further refine the SHC Gait Quality Index model based on state-of-the-art deep learning techniques and employ several validation strategies to assess efficacy statistically. His team will leverage this metric and related reduced dimensionality representations to predict post-operative outcomes – particularly by drawing direct comparisons with more traditional outcomes prediction. After development and testing of sitespecific models, across-site validation methods will be applied to the models for prediction of surgical outcomes. Comparisons of site-specific and across-site prediction models will allow an assessment of generalizability across sites. The team will also leverage these predictive models to assess various surgical options and rank order for clinical consideration. Furthermore, features which the model uses for prediction will be compared to metrics used clinically, with discrepancies between clinical and model use explored. Finally, this entire process has been established for surgical outcome prediction, and Dr. Albert’s team will adapt the process for similar evaluation of rehabilitation approaches – which similarly have highly variant patient populations and clinical practices necessitating dimensionality reduction strategies for generalizability and inference.
Discussion: No discussion notes