Identification of Schizophrenia from other Major Psychiatric Disorders using Machine Learning Techniques

Read full text Summarise with VIRAT
Authors: Aman Kapoor, Vishal Meena, Sandhir Kumar, Madhu Kumari, Shalini Mahato
Cited by: 0
References: 9

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Schizophrenia is a chronic mental illness, which is characterized by impairment of thoughts, percepts, affect, and behaviors, as a result of which there is significant functional impairment.The standard diagnosis for this condition is greatly dependent upon clinical evaluation, which is inherently time-consuming, subjective, as well as poor in distinguishing between schizophrenia and other chronic mental illnesses.Though EEG analysis has been recognized as a potential marker, previous studies have been criticized for employing small samples, moderate level of correctness in classification, as well as poor coverage of disorders.This paper proposes to examine whether an EEG-based solution using machine learning exists for this issue.A publicly available rest state EEG database of 945 individuals, including 850 patients diagnosed with six types of mental illness, as well as 95 controls, is used in this study.Six types of machine learning classifiers, namely Decision Tree, Random Forest, Support Vector Machine, Gradient Boosting, AdaBoost, and XGBoost, have been used to examine which ones work in terms of correct classification of results.The results of gradient boosters in classifying schizophrenia from normal controls is 82.98%, schizophrenia from mood disorders of 85.98%, schizophrenia from obsessive-compulsive disorder of 89.36%, but only 53.19% for schizophrenia from trauma and stress-related disorders, respectively, showed that AdaBoost performed either equal to, or better than, gradient boosters in several instances, including schizophrenia vs. obsessive-compulsive disorder at 91.49%.The results support EEG as an actual potential marker for schizophrenia diagnosis, but they also support nomological validity issues in current methods, which require better features, better algorithms, and more data to make them useful for diagnosis.

Fields of Study & Subjects

EEG and Brain-Computer Interfaces Schizophrenia research and treatment Emotion and Mood Recognition

Publication details

Journal / source
Advances in engineering research/Advances in Engineering Research
Pages
369–376
Publication date
30 March 2026
Language
EN
Document type
Book chapter