EEG Signal Analysis for Automatic Detection of Psychiatric Diseases

EasyChair Preprint 11657

Abstract

Neurological diseases such as epilepsy, Parkinson’s disease, and Alzheimer, or psychiatric diseases like depression, personality disorders, schizophrenia, or addictive behavior, among many others, affect numerous people around the world. Diagnosing these diseases is a challenge in medicine. The symptoms of neurological and psychiatric diseases can vary greatly, making it difficult for healthcare professionals to accurately diagnose and treat patients. Hence, it is based on the interpretation of symptoms by doctors and the analysis of EEG signals (the electroencephalogram).

There is therefore an important need for diagnostic support systems to assist doctors in their decision-making. Particularly in Algeria, where the health system suffers from a shortage of doctors specializing in neurology and psychiatry.

In this work, we compare the performance of different methods used in creating systems to assist in the medical diagnosis of an EEG signal in the case of a mental illness. This system will be designed using intelligent algorithms on electroencephalogram signals (EEGs), which are generally non-stable and complex and whose interpretation is long and laborious. Hence, the application of artificial learning algorithms such as random forest (RF) or deep neuronal networks.

This study brings light to the importance of machine learning algorithms in significantly reducing the time and effort required for interpreting EEG signals. We also raised the critical short-come of these same algorithms under some conditions in real-world problems, such as imbalanced datasets.

Keyphrases: EEG analysis, Psychiatric Disorder, automatic diagnosis, machine learning, neurological disease

@booklet{EasyChair:11657,
  author    = {Saidi Meryem},
  title     = {EEG Signal Analysis for Automatic Detection of Psychiatric Diseases},
  howpublished = {EasyChair Preprint 11657},
  year      = {EasyChair, 2024}}