SGER: Exploratory Research to Diagnose Dementia from Vehicle Operation Data

Yale University, New Haven CT

Investigators

Abstract

0302255 Kuc This research explores the feasibility of diagnosing and monitoring an operator's state of dementia by characterizing on-road operation of a vehicle. Five elements comprise the research plan: the human transfer function, vehicle operation data, dementia modeling, measures extracted from data that correlate with dementia model parameter values, and an experimental means to assess the performance of the methods. The experiments involve normal volunteers (engineering undergraduates who assist in the conduct of the research) drive an electric scooter equipped with sensors to monitor vehicle location, speed, acceleration, and direction; position of nearby objects; and, the driver's actions in controlling the scooter joystick. The scooter can also be programmed to modify the driver's actions, as if the driver had dementia. Signal processing procedures extract features in the driving data that correlate with dementia, which include slowed reaction speed, poor judgment and inappropriate actions. In these experiments the "correct answer" is known because the "dementia simulator" mimics slowed reaction speed by introducing a delay in the joystick control. Similar random modifications in the joystick signals simulate poor judgment and inappropriate actions. The vehicle then illustrates the degradation of driving behavior as the degree of dementia increases. Preliminary scooter results and computer simulations verify this claim. Results of the research are expected to be reported in the IEEE Transactions on Biomedical Engineering and the IEEE Transactions on Robotics and Automation for the engineering community, and the Journal of the American Geriatrics Society for the clinical community.