A comparative evaluation of EEG-based deep learning models for schizophrenia detection with cross-dataset validation and explainable AI

Key Points

• This research aims to evaluate the effectiveness of EEG-based deep learning models for diagnosing schizophrenia through cross-dataset validation.
• Utilized time-frequency analysis in conjunction with deep learning for EEG data processing.
• Implemented CNN architectures for spectral feature extraction from EEG signals.
• Analyzed interpretability of models in relation to neurophysiological patterns associated with schizophrenia.
• Showed improved detection accuracy for EEG-based models using spectral features; specifics not quantified.
• Provided insights correlating model predictions with known neurophysiological markers in schizophrenia.
• Emphasized the importance of explainability in AI models for clinical applications.