Secondary Logo

Institutional members access full text with Ovid®

Machine Learning in Modeling High School Sport Concussion Symptom Resolve

Bergeron, Michael F.1; Landset, Sara2; Maugans, Todd A.3; Williams, Vernon B.4; Collins, Christy L.5; Wasserman, Erin B.5; Khoshgoftaar, Taghi M.2

Medicine & Science in Sports & Exercise: January 25, 2019 - Volume Publish Ahead of Print - Issue - p
doi: 10.1249/MSS.0000000000001903
Original Investigation: PDF Only

Introduction Concussion prevalence in Sport is well-recognized; so too is the challenge of clinical and return-to-play management for an injury with an inherent indeterminant time course of resolve. Clear, valid insight to the anticipated resolution time could assist in planning treatment intervention.

Purpose This study implemented a supervised machine learning-based approach in modeling estimated symptom resolve time in high school athletes who incurred a concussion during sport activity.

Methods We examined the efficacy of 10 classification algorithms using machine learning for prediction of symptom resolution time (within seven, fourteen, or twenty-eight days), with a dataset representing three years of concussions suffered by high school student-athletes in football (most concussion incidents) and other contact sports.

Results The most prevalent sport-related concussion reported symptom was headache (94.9%), followed by dizziness (74.3%) and difficulty concentrating (61.1%). For all three category thresholds of predicted symptom resolution time, single-factor ANOVAs revealed statistically significant performance differences across the ten classification models for all learners at a 95% confidence level (P=0.000). Naïve Bayes and Random Forest with either 100 or 500 trees were the top-performing learners with an area under the ROC curve performance ranging between 0.666 and 0.742 (0.0-1.0 scale).

Conclusions Considering the limitations of these data specific to symptom presentation and resolve, supervised machine learning demonstrated efficacy, while warranting further exploration, in developing symptom-based prediction models for practical estimation of sport-related concussion recovery in enhancing clinical decision support.

1SIVOTEC Analytics, Boca Raton, FL;

2Department of Computer and Electrical Engineering and Computer Science, Florida Atlantic University, Boca Raton, FL;

3Nemours Children’s Hospital, Division of Neurosurgery, Orlando, FL;

4Kerlan-Jobe Center for Sports Neurology, Los Angeles, CA;

5Datalys Center for Sports Injury Research and Prevention, Inc., Indianapolis, IN

Corresponding author: Michael F. Bergeron, Ph.D., FACSM, SIVOTEC Analytics, Boca Raton Innovation Campus, 4800 T-Rex Avenue, Suite 315, Boca Raton, FL 33431, e-mail:, (706) 284-7142

The NATION project was funded by the National Athletic Trainers’ Association Research & Education Foundation and Central Indiana Corporate Partnership Foundation in cooperation with BioCrossroads. Content of this report is solely the responsibility of the authors and does not necessarily reflect the views of any of the funding organizations. The authors declare that the results of this study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine.

Accepted for Publication: 9 January 2019

© 2019 American College of Sports Medicine