Multimodal Identification of Depressive Symptoms in the Elderly

Multimodal Identification of Depressive Symptoms

Screening with depression scales alone is subjective, and relying on single-modal data often leads to incomplete identification of symptoms that are easily missed or misdiagnosed. In this study, we first aim to use artificial intelligence to construct a depression symptom recognition model, concatenate multimodal features such as facial expression, audio, text, and postural behavior, and deeply fuse them to construct a multimodal model.

Inclusion Criteria:

• Elderly individuals aged 60 years or older.
• Residents of the community (with residence time of more than 6 months).
• Possessing certain reading, writing and comprehension skills, being able to communicate with researchers without obstacles, and being able to independently complete the measurement of various indicators or, although unable to independently fill out the questionnaire, being able to independently make evaluations of the questionnaire items.
• Those who have given informed consent.

Exclusion Criteria:

• Those who meet the diagnostic criteria for cognitive impairment (dementia) as stipulated in DSM-5, and/or who suffer from severe physical diseases (such as advanced cancer, cardiovascular and cerebrovascular diseases, etc.).
• Those who are undergoing antidepressant treatment.
• Elderly individuals who have had suicidal thoughts in any psychological assessment, should be referred to the psychological hotline platform in Wuhan.