Metaheuristic Optimization Review MOR 3066-280X 10.54216/MOR https://www.americaspg.com/journals/show/3856 2024 2024 Faculty of Engineering, Technology and Built Environment, UCSI University, Kuala Lumpur 56000, Malaysia Wei Wei In this case, the diagnostic and statistical manual for mental disorders has experienced increased advancements in deep generative models (DGMs) that incorporate deep learning in analyzing neuroimaging information. The following review looks at different approaches that have been used in the classification of mental disorders and the specific performance of DGMs like GANs and VAEs. In classifying psychiatric symptoms, it remains challenging to represent the inherent intricacy of data by conventional methods. Thus, techniques that are more accurate are needed to identify complex patterns in extensive data. The newer studies also suggest that DGMs yield higher accuracy than traditional machine learning approaches because the most important features can be identified without requiring significant feature engineering. For example, using GANs to distinguish between major depressive disorder and healthy controls surpasses traditional classifier accuracy by remarkable margins. Moreover, this review contrasts the DGM architectures and their implementations in various psychiatric disorders that can improve diagnostic accuracy and pathophysiological features of diseases. Altogether, the results of the present study emphasize the possibilities of DGMs’ contribution to the field of psychiatry and open possibilities for further studies to deliver more precise diagnostic classifications and enhance the efficacy of treatment by employing the perspective of personalized medicine. 2025 2025 42 52 10.54216/MOR.040205 https://www.americaspg.com/articleinfo/41/show/3856