Heart rate variability analysis for prediction of sleep onset in car drivers

The aim of our study is to identify parameters able to predict the onset of falling asleep. Firstly we have considered the balance of low frequency versus high frequency in PSD. A set of experiments demonstrates that, when a person tries to resist falling asleep, the LF/HF ratio of PSD computed from the HRV signal increases significantly a few minutes before becoming significantly lower during sleep state. Like a reaction to falling asleep, it causes high activity of the sympathetic system while the parasympathetic system decreases its activity: we can call this phenomenon a micro arousal. There are several methods for performing predictions. We use fuzzy decision logic to model the oncoming onset of sleep. The whole system consists of a signal acquisition and preprocessing subsystem, a feature extraction subsystem, and a fuzzy logic decision module to predict.