Psychiatric disorders are thought to develop and progress through complex interactions between an individual's innate traits and temporary psychological states caused by stress and environmental changes. However, the relationship between these two factors and the molecular basis driving the transition of mental states remain unclear.In this study, we aim to comprehensively characterize the “state” common to psychiatric disorder by integrating diverse hierarchical data, including co-culture experiments using neurons and microglia-like cells derived from patient-derived iPS cells, multi-omics analysis, brain imaging data, and behavioral measurements.In particular, this study will develop a unique machine learning method to structure states that have not been classified to date and extract the biological and psychological factors driving their transitions. This is expected to enable the identification of new biomarkers for disease progression prediction and advance our understanding of the pathophysiology of psychiatric disorder.