National Institute of Mental Health 1R01MH140004-01A1

Enhancing Suicide Risk Detection through Computer Vision: a Novel Approach to Tissue Damage Analysis in Emergency Care

A research project funded by the National Institute of Mental Health to develop and validate computer vision and deep learning methods for analyzing tissue damage images to improve suicide risk detection in emergency care settings, with a focus on implementation and bias mitigation.

Description

This project aims to use computer vision technology to analyze images of self-injury tissue damage from youth and young adults in emergency departments to improve the identification of those at suicide risk. It will also provide insights into potential barriers and facilitators of this technology’s implementation in healthcare settings. Building on prior research, the study will expand the application of computer vision techniques to predict prospective suicide attempt (SA) risk more accurately by analyzing images of tissue damage for self-injury presence and severity. Participants aged 12 to 25 presenting with psychiatric concerns at MGB EDs will have standardized images of their arms taken at baseline. Follow-up assessments at 1 and 6 months will include remote surveys and medical record reviews to evaluate the predictive utility of signals derived from skin images. Deep learning techniques will be used to detect and classify tissue damage indicators of suicide risk. The study will also evaluate model performance across racial and ethnic minority groups to mitigate bias. Additionally, the research will explore implementation factors related to image collection and algorithm integration into electronic health records (EHRs). Successful completion will demonstrate the utility of computer vision at the point of care and inform future scale-up efforts. This approach aims to enhance objective suicide risk assessment and monitoring, aligning with NIH priorities for innovative mental health tools.

View original record