A Fast-Running, End-to-End Concussion Risk Model for Assessment of Complex Human Head Kinematics

Abstract An end-to-end, mechanism-based concussion risk model, linking head motion to axonal injury, has been demonstrated to predict concussion outcomes with greater sensitivity and specificity than external correlates such as peak head acceleration. The development of this model was driven by the need to more accurately translate head-worn sensor measurements into injury assessment in near-real time. The full end-to-end model is a detailed multi-scale model, composed of complex components (e.g., a human head finite element model), is computationally expensive, and requires specialized software. For practicality, this research-level model must be simplified into a standalone, fast-running algorithm that can be embedded on the microprocessor of a head-worn sensor. This article describes the development of a simplified, fast-running algorithm that delivers comparable results to those of the full end-to-end model. The dynamic axonal response of the human head finite element model to head motion is mathematically modeled using a lumped parameter system fitted to the finite element model response for a range of head motions. The other component models of the full end-to-end model were similarly reduced. For the same head kinematic scenarios, the probabilities of concussion obtained from the end-to-end model and from the simplified algorithm are compared well.