Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : September 13, 2020 - September 16, 2020
Advanced automatic collision notification systems (AACN) enable early notification of road crashes as well as providing a prediction of occupant injury severity in road crashes. Occupant injury severity, when transmitted quickly and automatically, can be used by emergency medical services (EMS). This information allows EMS to respond quicker and be better informed and prepared when arriving at a road crash. Accurate injury prediction algorithms can improve survival rates of injured occupants. This is particularly important in Australia, as occupant injury severity in road crashes are higher than in Japan. In this study, the most appropriate injury risk factors that were found to influence occupant injury severity in South Australian road crashes were used. Combinations of the various risk factors produced around 33,000 injury prediction models. Each of these models was evaluated using Akaike's Information Criteria (AIC) to determine the optimal risk factor. The determined risk factor was used to construct a neural network injury prediction algorithm. The prediction accuracy of the algorithm was evaluated by 10-fold cross-validation. As a result of this study, an optimal injury prediction algorithm with high prediction accuracy according to AIC has been developed and is expected to be applied to AACN systems.
