Electronic Health Records to Detect Psychosis Risk

Improving outcomes in psychosis is reliant on efficient early detection of individuals at risk; however, current strategies are suboptimal. Electronic health records (EHRs) contain detailed individual-level data that are routinely collected as part of clinical care, providing a unique opportunity for personalised prognostication. This chapter explores current detection strategies, their limitations and how prognostic models leveraging EHR data have attempted to address them. Research in primary care has highlighted potential symptom-based predictors for future prognostic models and shown that frequency of consultations increases closer to psychosis onset. A transdiagnostic risk calculator in secondary mental health care has displayed adequate prognostic accuracy in different settings in the UK and US. It is the only prognostic model in psychiatry to be implemented in real-world clinical practice, showing good evidence of feasibility. Dynamic prognostic models may be better able to model the time course of psychosis risk compared to static models.