ConcussionWatch: A real-time method for concussion detection in athletes

Each year, athletes across the world are injured or killed from 3.8 million sports-related concussions, primarily related to contact helmet sports like football and lacrosse. Despite product development efforts at measuring hit-detection, determining cause of concussion, and predicting recovery, there is a lack of focus on real-time concussion detection for athletes. This study involved the construction of an electronic sensor containing device capable of being placed inside a lacrosse helmet. Proof-of-principle testing demonstrated that the device did not degrade comfort of wear, and was robust enough to withstand limited impacts as would occur during concussion causing hits during games. The electronic sensor stored a feedforward artificial intelligence, neural network model to detect real-time concussions. Uploading and running the neural network on/from the device ensured that the concussion-detecting and measuring functionality of the device was operational without reliance on Bluetooth® or on WiFi. The ML model was based on an online dataset containing actual concussion data correlated with the Head Injury Criterion (HIC) and the Gadd Severity Index (GSI); which in turn, was linearly interpolated with an equal amount of synthetic data. The model returned scores > 0.99 (for a binary decision of concussion/no-concussion) across metrics consisting of accuracy, recall, precision and F1, when tested on hold-out data. Additionally, a mobile app was coded to work in conjunction with the concussion-detection and measurement sensor by recording data that could be viewed by players, coaches, and medical professionals. Constructing the circuit with commercially available and marked-up parts led to a significantly cheaper device than currently available commercial devices; and one that was additionally equipped with real-time concussion detection functionality. The results of the study pose a potential advancement in the development of real-time concussion-detection technology and the ability to protect athletes against brain injury.