Schizophrenia is a severe mental disorder that significantly affects individuals quality of life. Therefore, early detection and precise diagnosis are crucial for effective intervention. However, due to the complex pathophysiology of schizophrenia, traditional methods fail to perform accurate and efficient diagnosis. Machine learning (ML) has proven to be a useful tool for handling complex data and detecting subtle changes in the brain for early detection, addressing the current challenges in psychiatry. This paper explores different data types, such as EEG, MRI, fMRI, behavioral, and genetic data, and ML models for schizophrenia diagnosis by examining specific data processing, feature selection, and classification methods. Gaps and challenges in current models, including data biases and model explainability, are also addressed in later sections. EEG data trained with supervised and deep learning models achieved the highest classification accuracy out of all the data modalities and ML models. The studies also found that functional connectivity, behavioral data, specific genes, immune-based metabolites and cytokines are all important biomarkers of SZ. This review aims to provide insight into the application of ML in the diagnosis of schizophrenia, highlighting its potential for broader clinical use.
Zijia Li (Thu,) studied this question.