Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


30563.pdf (3.44 MB)

ETD Abstract Container

Abstract Header

Machine Learning for the Prevention and Prognosis of Pediatric Head Injury in Sport

Richards, Nathaniel L
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1535633637793776

Abstract Details

2018, MS, University of Cincinnati, Engineering and Applied Science: Aerospace Engineering.
Researchers in sports medicine seek to reduce the overall occurrence and severity of collision-related injuries – specifically concussions. Physicians also seek to provide concussion patients with proactive care, based on self-reported symptoms following the head injury. Models that contribute to the prevention and prognosis of head injuries are desired by physicians, researchers, and trainers to provide predictions and insights for decision-support tools and training interventions. Athletes can be trained to improve their perception of the dynamic sport environment; targeted interventions have the potential to reduce player collisions by intervening on modifiable perceptual-motor variables. This work explores such a training intervention via virtual-reality environment, where athletes have full range of motion to perform sport-specific tasks. The equipment involved records each athlete’s perceptual-motor behaviors within training trials. Machine learning models are trained to predict trial performance outcomes based on these in-trial behaviors. The output predictions are then used to adjust the virtual environment, such that the athlete’s collision anticipation and avoidance improves. The resulting models are able to predict the time-to-goal of an athlete navigating to a waypoint in a dynamic, uncertain environment. Predictive models can also be constructed to aid physicians in providing proactive care for concussion patients that exhibit prolonged symptom recovery. Patient data from electronic medical records was used to train a variety of machine learning models. The original problem iiinvolved training regression models to predict concussion recovery times, given patient self-reported symptoms. The regression problem was reframed as a classification problem as a result of the failure of the regression algorithms to generalize. The challenges associated with the small sample size, noisy features, and self-reported symptoms are also explored. This work explores the successes and challenges in the development of precision medicine approaches based on machine learning for the prevention and prognosis aspects of pediatric collision-related injury in sport. Throughout this work, traditional regression and classi- fication models are compared to modern deep learning models. State-of-the-art training methods are explored and implemented in constructing deep neural network models. The resulting performance and design effort are then compared to that of traditional regression and classification models. Exhaustive hyperparameter searches are conducted to yield the models that exhibit the best generalization performance.
Kelly Cohen, Ph.D. (Committee Chair)
Adam Kiefer (Committee Member)
Ali Minai, Ph.D. (Committee Member)
Paula Silva, Ph.D. (Committee Member)
95 p.
Artificial Intelligence
sports medicine; machine learning; neural network; concussion; behavioral psychology

Recommended Citations

Citations

  • Richards, N. L. (2018). Machine Learning for the Prevention and Prognosis of Pediatric Head Injury in Sport [Master's thesis, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1535633637793776

    APA Style (7th edition)

  • Richards, Nathaniel. Machine Learning for the Prevention and Prognosis of Pediatric Head Injury in Sport. 2018. University of Cincinnati, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1535633637793776.

    MLA Style (8th edition)

  • Richards, Nathaniel. "Machine Learning for the Prevention and Prognosis of Pediatric Head Injury in Sport." Master's thesis, University of Cincinnati, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1535633637793776

    Chicago Manual of Style (17th edition)

ucin1535633637793776
360
© 2018, some rights reserved.Creative Commons License Machine Learning for the Prevention and Prognosis of Pediatric Head Injury in Sport by Nathaniel L Richards is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by University of Cincinnati and OhioLINK.