Bipolar disorder (BD) is a serious, complicated, familial aggregation onset of mental illness, which has the characteristics of five-low and one-high, namely high prevalence, high recurrence rate, high morbidity and mortality, high comorbidity rate and younger age characteristics. This situation will seriously influence one's behaviour or thinking, cognitive, emotional, social and occupational function, causing the heavy burden of disease. But, early recognition and early diagnosis are difficult to achieve at present. Based on the preliminary research results of the project team, it is found that BD can be identified early through specific dimensions, and early recognition is crucial for the prognosis of patients. The earlier the intervention for BD is implemented, the better the prognosis, especially the functional prognosis, but the difficulty lies in how to implement it. Establishing a high-quality clinical cohort of BD high-risk population is a necessary prerequisite. This study intends to establish a high-quality, large-sample cohort through multi-center, long-term and prospective cohort design and enroll 100 BD high-risk patients every year, a total of 400 cases in 4 years. The electronic mental health service platform will be used for ten years of intensive follow-up. Multi-modal data including clinical characteristics, genetic, cognitive, neuroimaging, sleep monitoring, eeg, eye movement, speech, facial expression and movement were collected to construct the database. On this basis, the interaction of biological factors, clinical risk factors, and environmental risk factors in the onset of BD is discussed to establish a big data prediction model for BD onset in high-risk populations. The effective subgroups of early intervention were analyzed and screened. An ethical and individualized prediction model of the effectiveness and safety of early intervention for the BD high-risk population was constructed. It is hoped that the smooth implementation of this project can provide empirical evidence for the early identification, prevention and intervention of BD. To provide clinicians with real data-driven decision-making guidance to assist in selecting personalized and precise treatment; Ultimately promote the prognosis and functional recovery of BD patients.