A Hybrid Neural Classifier for Depression Screening Using Electroencephalograms (EEG)

Abstract

Depression seems to be a psychiatric condition marked by recurrent episodes of depression. It is very crucial to realise that people of all ages in different parts of the world are being seriously affected by this condition. The early detection of this sickness will help to save many lives because it is now recognised as a global problem. Electroencephalogram (EEG) signals can be used to diagnose this mental disease since they can be analyzed to reveal the patients' current mental state. The benefits of a completely automated depressive detection scheme are discussed in detail in this study since manual analysis of the EEG data is time-consuming, laborious, and takes a great deal of expertise. This study introduces a new Electroencephalography computer-aided hybrid computational model for depression screening. The suggested approach makes use of windowing and long-short term memory (LSTM) architectures for sequence learning as well as CNN for temporal learning. The EEG data used in this model were acquired using neuroscan from some drug-free, symptomatic depressive patients, as well as few healthy individuals. The windowing approach is used by the model, which reduces calculation complexity and saves time. It gives more than 99 percent accuracy. The findings demonstrate the usefulness and accuracy of the hybrid CNNLSTM model for identifying depression in EEG data.