Large Language Models
2d ago
MIT researchers develop tool to assess suicide risk from text analysis
Sep 24, 2026
AI Summary
Scientists at MIT have created a language-processing tool that evaluates text for indicators of suicide risk. This tool, developed by a team led by Daniel Low, analyzes conversations from the Crisis Text Line to identify key risk factors and improve mental health crisis interventions.
- The tool aims to help counselors identify individuals at high risk of suicide during mental health crises by analyzing their language for specific indicators.
- Developed by Daniel Low and Satra Ghosh at MIT, the tool uses a lexicon of words and phrases linked to 49 suicide risk factors to estimate an individual's risk based on text conversations.
- The research involved analyzing de-identified texts from approximately 16,000 conversations with Crisis Text Line counselors, categorizing them into different risk levels: non-suicidal, suicidal ideation without imminent risk, and imminent risk.
- The study found that certain expressions, such as mentions of lethal means and substance use, were strong predictors of imminent risk, while traditional indicators like depression were less predictive in some cases.
- The predictive model assigns weights to risk factors based on their contribution to overall risk, allowing for accurate assessments of new conversations.
- Limitations of the lexicon include a lack of contextual understanding, but the model is designed to be interpretable and can be run on personal computers, making it accessible and cost-effective.
- The researchers emphasize the importance of human oversight in using the tool and stress the need for thorough validation before clinical application.
- The suicide risk lexicon and the software developed for it are being shared widely to assist in building similar tools for other mental health conditions.
suicide risklanguage processingnatural languageinterventionhealthcare