EEG Schizophrenia Classification with Comparison of Three Machine Learning Algorithms

General Background: Schizophrenia is a chronic mental disorder affecting millions globally, requiring improved diagnostic methods. Specific Background: EEG signals have emerged as promising biomarkers for schizophrenia classification through machine learning. Knowledge Gap: Despite prior advances, no systematic comparison of key machine learning algorithms—Logistic Regression, Random Forest, and Decision Tree—using EEG data for schizophrenia classification has been conducted. Aims: This study aims to compare the performance of these three algorithms in classifying schizophrenia from EEG signals using a dataset of 1932 samples. Results: Random Forest achieved the highest classification accuracy (86%) and AUC (0.912), outperforming Logistic Regression (accuracy 82%, AUC 0.865) and Decision Tree (accuracy 81%, AUC 0.871). Novelty: Unlike previous studies, this research provides a comprehensive algorithmic comparison using EEG-derived features, integrating feature importance, calibration, learning curves, and statistical tests. Implications: The findings establish Random Forest as a robust classifier for EEG-based schizophrenia detection, offering a foundation for developing clinically relevant, cloud-based diagnostic support tools that can facilitate early detection and personalized treatment planning in mental health care.Highlight : Random Forest achieved the highest accuracy and AUC in schizophrenia classification. EEG data were processed using STFT, Wavelet Transform, and Band Power features. Comparison of three algorithms offers a systematic basis for clinical application.