Detecting Concussion History in Athletes Using Pose Estimation and Machine Learning

Abstract

   Concussions present a significant risk to athletes, with females exhibiting higher rates and prolonged recovery times than males. Current sideline concussion detection methods, such as the King-Devick test, suffer from validity issues, especially among young athletes, highlighting the need for more accurate and objective assessment tools. This study investigates the feasibility of using pose estimation technology, specifically Microsoft Kinect V2, to assess postural stability in varsity athletes with a concussion history. Inspired by previous research utilizing force plates, our study analyzes video recordings of athletes performing specific exercises to detect dynamic balance deficits. In a cross-sectional study of 444 varsity athletes in 2022 and 464 varsity athletes in 2023, results reveal significant differences in movement mechanics between concussed and control groups, with the Drop Vertical Jump (DVJ) exercise demonstrating the highest discriminatory power. Notably, concussed individuals exhibit longer time to stabilization (mean difference = 0.089 seconds, p = 0.046) during DVJ, indicating potential lingering balance impairments. While single leg squat and single leg hop exercises showed fewer discriminatory metrics than DVJ, they still provide valuable insights into balance capabilities. Our research also studied 20 varsity athletes who sustained one or many concussions from 2022 to 2023 and compared their DVJ exercise metrics before and after injury. The results of this prospective study demonstrated no statistically significant differences between their 2022 and 2023 jumps for all computed metrics, suggesting that the occurrence of concussion(s) did not have a measurable impact on the athletes' jumping mechanics or dynamic balance for the DVJ over this period.