Establishing a predictive model for unsafe driving in the elderly using artificial intelligence and elucidating the neural basis

Abstract

In Japan, there is a social issue associated with accidents caused by elderly drivers. Although mandatory cognitive function tests when renewing driving lessons screen for dementia, they cannot account for risk of unsafe driving due to the natural ageing process. In his work, Associate Professor Bun Yamagata, Department of Neuropsychiatry, Keio University School of Medicine, has been using imaging techniques such as MRI and NIRS and combining the data obtained with neuropsychological evaluations and artificial intelligence (AI) technology to better understand brain function and structural abnormalities within the brain. In partnership with ergonomics and mechanical engineering specialist Associate Professor Motoki Shino, from The University of Tokyo, he is conducting an innovative study on driving behaviours and brain atrophy patterns. By combining engineering and neuropsychology techniques the researchers will develop algorithms that predict the risk of dangerous driving behaviours in elderly drivers. In research exploring the relationship between the structural connectivity of white matter in the brain and the driving ability of healthy older people Yamagata found that changes in the white matter within the dorsal attention network may contribute to a higher risk of unsafe driving behaviours in healthy elderly people. The researchers have used machine learning to develop a model that predicts the risk of unsafe driving in healthy older people with high accuracy from neuropsychological tests and practical visual acuity.