AI-Based Identification of Head Impact Locations, Speeds, and Force Based on Head Kinematics Simulations.

Abstract

With the development of wearable sensors, head kinematics data have become widely available. However, key impact information-such as impact direction, speed, and force-which is crucial for helmet development, is still not being directly measured. This study presents a deep learning model designed to accurately predict these head impact parameters from head kinematics during helmeted impacts. Leveraging a dataset of 16,000 simulated helmeted head impacts using the Riddell helmet finite element model, we implemented a Long Short-Term Memory (LSTM) network to process the head kinematics: linear accelerations and angular velocities. In the simulated dataset, the models accurately predict the impact information describing impact direction, speed, and the impact force profile with $R^{2}$ exceeding 70% for all tasks. Further validation was conducted using an on-field dataset recorded by instrumented mouthguards and videos, consisting of 79 head impacts in which the impact location can be clearly identified. The deep learning model significantly outperformed existing methods, achieving a 79.7% accuracy in identifying impact locations, compared to lower accuracies with traditional methods (the highest accuracy of existing methods is 49.4%). The precision on simulations underscores the model's potential in enhancing helmet design and safety in sports by providing more accurate impact data. Future studies should test the models across various helmets and sports on large in vivo datasets to validate the accuracy of the models, employing techniques like transfer learning to broaden its effectiveness.