Optimal EEG Segmentation for Microsleep Detection Based on Machine Learning

Abstract Paroxysmal brain state changes, such as microsleep events in drivers, are presumably subcortically induced and accompanied by cortical processes. This raises questions of how stable persistent the processes that can be observed with EEG real time. For this purpose, recordings four night-driving simulation studies including 79 subjects used to analyze large time window temporal offset segment must achieve maximum classification rate. From each segment, power spectral densities were estimated using modified periodogram method averaged narrow bands. They then processed gradient boosting machines order map them one two types: or sustained attention. Segment length found have moderate dramatic effects on recognition rate, respectively.