According to the World Health Organization, more than 300 million people worldwide suffer from depression. 60% of adults with mental illness do not receive any mental health services; social stigma, cost, and access to services all get in the way of receiving treatment. One key to effective treatment for depression is early and accurate diagnosis.
Researchers at Stanford University have published a paper describing a system that can automatically detect symptoms of depression simply through spoken language analysis and facial recognition. They applied a number of different artificial intelligence (AI) machine learning tools to the data recorded in patient interviews with a computer avatar.
Compared with the widely-used Patient Health Questionnaire (PHQ), the AI system managed just a 15% average error. It achieved 83.3% sensitivity and 82.6% specificity, which would make it a useful tool for at least screening patients. The researchers’ goal is to make this system available within a smartphone, since it is capable of recording audio and video. A computer-administered interview could provide data that would be analyzed in the smartphone and reports forwarded to the individual’s healthcare provider.
Such a system could greatly reduce the cost and anxiety related to mental health screenings, while making it broadly available to individuals worldwide.