Transformer models analyze EEG data to detect schizophrenia
Schizophrenia Detection from EEG Signals: A Transformer Framework with Spectrogram Representation
Artificial Intelligence
Summary
Schizophrenia is a mental disorder that is currently diagnosed mostly through doctor assessments. The authors used brainwave recordings called EEGs and converted them into images showing frequencies over time. Then, they used special computer models called transformers combined with other machine learning methods to identify patterns linked to schizophrenia. Their method showed good accuracy by carefully testing on data from different people.
What this means in practice
- •For clinical diagnostic tool developers: Develop automated EEG-based diagnostic aids for schizophrenia screening using time-frequency spectrograms analyzed by transformer-based models.$Commercial implications: Enables sale of AI-powered medical diagnostic software to hospitals and clinics for mental health assessments.
- •For medical device engineers: Integrate transformer frameworks with EEG hardware for real-time schizophrenia detection supporting clinical decision making.
Authors
Abtin Shafiei, Mohsen Hooshmand, Majid Ramezani
Abstract
Schizophrenia is a serious psychiatric disorder that affects millions of people worldwide, and its diagnosis remains primarily dependent on clinical assessment. Electroencephalography (EEG) provides a non-invasive approach to investigate brain activity and has shown potential to support automated Schizophrenia detection. However, existing EEG-based classification studies often suffer from limitations including small datasets, inconsistent preprocessing strategies, and evaluation protocols that may not adequately prevent subject-related data leakage. In this study, we propose an EEG-based Schizophrenia classification framework that transforms preprocessed EEG recordings into time-frequency representations using the Short-Time Fourier Transform. The generated spectrogram images are classified using both conventional Machine Learning algorithms, including Support Vector Machines, Random Forests, and XGBoost, and Deep Learning models, including convolutional architectures and CNN-Transformer hybrids. To ensure reliable evaluation, all data partitions are performed at the subject level. Experimental results demonstrate that the proposed approach achieves competitive classification performance, with the CNN-Transformer (CT-SZ) model achieving an AUC-ROC of 88.41% and the CNN + Squeeze and Excitation + Transformer (CST-SZ) achieving an AUC-ROC of 92.88% on the independent test set.