8 - Customized deep learning algorithm for drowsiness detection using single-channel EEG signal

Abstract

An electroencephalogram (EEG) is a graphical tool used to analyze the brain's state. Drowsiness is one of the brain's crucial states, and it is the primary cause of accidents globally while driving vehicles. Mental sleepiness and obviousness are a portion of the explanations behind drowsiness. Therefore, identifying the drowsiness state of a person is an important task to avoid accidents. Manually analyzing an EEG signal with the naked eye is very difficult; hence an automatic system is required. In this chapter, a deep learning algorithm based on a convolutional neural network is proposed to identify the drowsiness state automatically. A single-channel raw EEG signal is used as input of the proposed deep learning algorithm in this work. The deep learning algorithm automatically extracts the different features from the applied EEG signal. The proposed deep learning classifier detected a drowsiness state with an accuracy of 98.20%. In the cross-subject validation process, the classifiers provide approximately almost the same efficiency of 98.52%.