Analysis of Parameters that Characterize Drowsiness Based on EEG, ECG and EOG Records

Drowsiness in drivers is one of the main causes of fatal road accidents. The development of early detection systems, which could also be incorporated into the vehicle, would help to prevent accidents. Sleepiness can be detected early through the processing and analysis of biological signals. In this research work, a preliminary study is proposed based on the qualitative analysis of time and spectral parameters extracted from EEG, ECG and EOG recordings of volunteers during the use of a driving simulator. The records of 10 subjects were processed, 14 features were extracted for each subject and each record, and 653 sleepiness events were studied. The analysis implied averaging features and the evaluation of their variation during the transition from the waking state to the drowsy state. For example, the spectral features FC, Q1F, Q3F, MaxF, FM, PF and RH/L of the EEG were found to increase. Likewise, it was observed that the time parameters did not show variation for any of the 3 signals. The results obtained showed the parameters that provide more information to the problem and can be included in a quantitative analysis, in order to generate an indicator that can be implemented in an advanced system for detecting drowsiness in drivers. In this way, fatal accidents caused by due to drowsiness could be prevented.