Pycaret for the evaluation of classification methods in order to set up a decision making system for the early diagnosis of schizophrenia by EEG

According to the World Health Organization, approximately 24 million people suffer from schizophrenia worldwide. However, there is no clinical examination to diagnose this pathology. Specialists rely on symptoms and signs. Which can be problematic. The aim of this work is to set up a Decision-Making System allowing the early diagnosis of schizophrenia by EEG. Pycaret is the classification tool used throughout this work. A python code was previously written for the preprocessing and analysis of EEGs. This last task led to obtaining an Excel file which is passed as input to pycaret. This method led to an overall accuracy of 60%. The confusion matrix obtained showed that approximately 67% of people who are actually ill are classified as such and 63% of those who are not ill are classified as not ill. However, this work can be improved by using ensemble methods and graph-based techniques for example.