Optimization in channel selection for EEG signal analysis of Sleep Disorder subjects

Abstract

Deep learning, a branch of artificial intelligence (AI), is establishing a very promising approach for electroencephalogram (EEG) signals to sense and extract features from raw signals. The presented work here focuses on optimization in channel selection and batch size for EEG signals to identify sleep disorder subjects from normal ones with the deep learning-based model. It is observed that the data from several electrodes reduced recall obtained also starts reducing. To implement this work, an openly available Physionet EEG dataset from various ten electrodes is used. The long short-term memory (LSTM) method from a class of Recurrent Neural Networks (RNN) is nominated for the detection of sleep disorder subjects. The suggested method can achieve a recall of 99.8% on the test dataset.