On the Feasibility of Real-Time Prediction of Driver's Traffic Accident Risk with Auto ML using Demographics and Brain Concentration Levels Information

Abstract

This paper presents a real-time prediction model for driver's traffic accident risk during the next minute using the information of the driver's concentration levels from the previous minute combined with the driver's demographic information. The predictive model is built using automated machine learning tools. The data used in this work was collected in a previous study that developed models for driver's traffic accident prediction during a 10 minutes driving simulation exercise in which the drivers received phone alert distractions.