Predicting and preventing risk of suicide with AI
A research team at University of Vanderbilt has developed a machine learning algorithm designed to predicting the risk of an individual attempting suicide in the next two years with a more than 80 % accuracy. The accuracy even rises to 92 % for a prediction on the following week. The algorithm was fed with medical data from 5,000 patients admitted in the university medical centre for self-harm signs or suicidal thoughts and those of 12,000 average patients.
Big data here comes to serve a digital intelligence capable of detecting signs of alert that could go unnoticed by the medical staff. This research work is a part of a series of innovations trying to predict human behaviour with digital intelligence (cf. Elderly Fall Prevention System, AI Couple Therapy Outcome Prediction, Cloudwalk).
Spotted in 2018
- United States