Automatic Micro-sleep Detection under Car-driving Simulation Environment\n using Night-sleep EEG

A micro-sleep is a short sleep that lasts from 1 to 30 secs. Its detection\nduring driving is crucial to prevent accidents that could claim a lot of\npeople's lives. Electroencephalogram (EEG) is suitable to detect micro-sleep\nbecause EEG was associated with consciousness and sleep. Deep learning showed\ngreat performance in recognizing brain states, but sufficient data should be\nneeded. However, collecting micro-sleep data during driving is inefficient and\nhas a high risk of obtaining poor data quality due to noisy driving situations.\nNight-sleep data at home is easier to collect than micro-sleep data during\ndriving. Therefore, we proposed a deep learning approach using night-sleep EEG\nto improve the performance of micro-sleep detection. We pre-trained the U-Net\nto classify the 5-class sleep stages using night-sleep EEG and used the sleep\nstages estimated by the U-Net to detect micro-sleep during driving. This\nimproved micro-sleep detection performance by about 30\\% compared to the\ntraditional approach. Our approach was based on the hypothesis that micro-sleep\ncorresponds to the early stage of non-rapid eye movement (NREM) sleep. We\nanalyzed EEG distribution during night-sleep and micro-sleep and found that\nmicro-sleep has a similar distribution to NREM sleep. Our results provide the\npossibility of similarity between micro-sleep and the early stage of NREM sleep\nand help prevent micro-sleep during driving.\n