A multi-sense deep learning platform and interview protocol used for detection and assessment of depression and schizophrenia.
Depression and schizophrenia are mental illnesses with significant economic impacts on society. In the US, studies have found that depression and schizophrenia costs can reach up to $210 and $155 billion USD annually, respectively. Current methods of diagnosis and evaluation rely on cost-prohibitive psychiatric observation of patient signs and symptoms, which are subject to high variabilities amongst individuals, complicating diagnosis.
This invention is an automated, cost-reductive approach that enables early-stage detection of schizophrenia and depression and its severity assessment across a wide range of individuals. This multi-sense platform utilizes a deep neural network to analyze visual and audio data collected remotely from patients to classify anxiety state. This platform is capable of predicting as well as assessing the severity of depression and schizophrenia based on facial emotion, speech pattern, vocal expression, and physiological data.
23 hours and 41 minutes of audio from interviews and 30 individuals have been utilized in the development of this software algorithm. Publication Cotes, R. O., Boazak, M., Griner, E., Jiang, Z., Kim, B., Bremer, W., . . . Clifford, G. D. (2022). Multimodal Assessment of Schizophrenia and Depression Utilizing Video, Acoustic, Locomotor, Electroencephalographic, and Heart Rate Technology: Protocol for an Observational Study. JMIR Res Protoc, 11(7), e36417. doi:10.2196/36417 Z. Jiang et al., "Multimodal Mental Health Digital Biomarker Analysis From Remote Interviews Using Facial, Vocal, Linguistic, and Cardiovascular Patterns," in IEEE Journal of Biomedical and Health Informatics, vol. 28, no. 3, pp. 1680-1691, March 2024, doi: 10.1109/JBHI.2024.3352075