Evaluation and prediction of diffuse axonal injury based on optimization strategy in vehicle collision accidents

The brain is one of the most critical parts of the human body, and it is vulnerable in vehicle collision accidents. Statistically, traumatic brain injuries (TBIs) account for about half of the 1.3 million deaths and 50 million injuries in annual road traffic accidents around the world. However, there are currently no universally accepted and specialized criteria for the different types of brain injuries, even though a series of injury criteria has been presented using mathematical combinations of kinematic parameters. To reduce TBIs and improve the safety performance of vehicles, we established a new brain injury index (BII) by maximizing the correlation between the kinematic parameters and strain-based measures such as cumulative strain damage measure (CSDM) and maximum principal strain (MPS), which employed 218 crash test data and the simulated injury monitor (SIMon) model from the National Highway Traffic Safety Administration website. In the process of establishing the BII, we combined the K-Nearest Neighbor with quadratic regression to enhance the correlation between the kinematic metrics and CSDM/MPS by eliminating the influence of some outlier data and used the genetic algorithm to obtain the optimal weight ratios of several kinematic parameters with strong correlations. The assessment capability of the proposed BII was more superior and reliable than other indexes when compared with 15 existing kinematic-based criteria. Finally, we developed a simple BII (SBII), which ignored the influence of the translational velocity and acceleration, and used it to establish three prediction models of brain injury based on artificial neural network learning, which achieved the quantitative description of the relationship between the kinematic parameters and CSDM/MPS.