Abstract
The surge in artificial intelligence (AI) applications within psychiatric research and diagnosis has witnessed significant growth in recent years. This study investigates the use of AI to facilitate early medical condition diagnosis and enhance our understanding of disease progression, particularly in the realm of psychiatric disorders. The primary objective is to explore and employ various AI algorithms for the identification of biomarkers associated with psychiatric conditions. Data and methods involve the application of diverse algorithms for classifying psychiatric disorders, with a meticulous comparison of their accuracy. Additionally, a model is developed based on these algorithms, aiming to optimize diagnostic precision. Results indicate a notable 70% accuracy in the dataset, highlighting the efficacy of deep learning approaches in handling extensive data sets. The findings underscore the potential of deep learning in clinical datasets and its application in the future detection of mental health issues. Despite the commendable performance of deep learning, criticisms persist regarding its accountability during development and assessment phases. While AI has made significant strides in detecting psychiatric diseases, this study identifies areas for improvement in AI-based applications. Notably, the current model's limited generalizability due to its analysis of homogeneous datasets prompts the consideration of future approaches, including migration learning, multi-view learning, and ensemble learning, to handle diverse and extensive psychiatric disease data sets.