Lancet Digital Health https://doi.org/10.1016/S2589-7500(20)30024-8 (2020)

Machine learning can be used to predict the first episode of psychosis in people with schizophrenia 1 year before its occurrence.

Schizophrenia is a disabling psychiatric disorder, and the first episode of psychosis defines long-term outcomes. There are interventions that can prevent this event and could improve outcomes if used at the right time.

Raket et al. used a complied set of data from diverse electronic health record databases to develop the DETECT (‘dynamic electronic health record detection’) neural network to predict the first event of psychosis in people with schizophrenia. The model should be refined in further population-based settings, but they found it had adequate prediction accuracy.