Estimating suicide risk from text

Scientists at MIT's McGovern Institute for Brain Research have developed a language-processing tool to identify individuals at high risk of suicide by analyzing the language they use during mental health crises. The tool, created by Daniel Low and Satra Ghosh, uses a custom-built list of words and phrases linked to 49 suicide risk factors to estimate an individual's risk level.

The researchers collaborated with the Crisis Text Line, a global mental health nonprofit, to analyze de-identified texts from approximately 16,000 conversations with crisis counselors. The tool accurately predicts suicide risk from text conversations with crisis counselors and could help with risk assessment in clinical settings and crisis-support situations.

Suicide attempts are difficult to predict, with dozens of risk factors interacting in complex ways. Factors such as psychiatric symptoms and disorders, environmental and social stressors, and personal circumstances can all contribute to suicidal thoughts and behaviors. The tool uses a machine learning model to search for specific words and phrases in a lexicon and predict risk levels.

The model assigns weights to each risk factor based on its contribution to risk, with mentions of lethal means for suicide being weighed heavily.