Researchers have adapted a tool for local use that could transform early mental health care by identifying young adults at risk of developing psychosis—before symptoms become severe.
Psychosis is a mental health condition in which individuals may lose touch with reality, experiencing hallucinations, delusions, or confusion. While these symptoms can be deeply distressing, early detection and intervention can significantly improve outcomes.
The ‘transdiagnostic risk calculator’ originally developed by Professor Paolo Fusae-Poli in Kings College London uses artificial intelligence to analyse electronic health records and detect those at risk of psychosis. By examining patterns such as changes in sleep, substance use, and emotional well-being, the tool gives clinicians a timely opportunity to intervene and provide tailored support.
A study, led by Professor Kathryn Greenwood from the University of Sussex and Sussex Partnership NHS Foundation Trust, and funded by the Applied Research Collaboration Kent, Surrey and Sussex, evaluated the tool using data from more than 63,000 anonymised patient records in South East England. With a predictive accuracy score of 0.71—on par with similar systems used elsewhere—the results indicate that this tool could become an important asset for mental health services, especially in regions where early intervention remains underutilised but urgently needed.
“Psychosis is a serious condition that can cause hallucinations, delusions, and confusion. Early intervention is critical, yet up to 95% of at-risk individuals are not identified by specialist services before their first episode,” explains Professor Greenwood. “Our research shows that by using existing clinical data and smart technology, we can change this. Adapting the calculator to local services has allowed us to create a practical tool that can help clinicians intervene earlier and offer the right types of much-needed support.”
The risk calculator employs machine learning and natural language processing to scan clinical notes for warning signs like sleep disturbances, substance use, and emotional distress. The adapted tool demonstrated strong accuracy and could be rolled out in community mental health settings.
This collaborative project involved the University of Sussex, Brighton and Sussex Medical School, Sussex Partnership NHS Foundation Trust, University of Oxford, Oxford Health NHS Foundation Trust, Akrivia Health, King’s College London, South London and Maudsley NHS Foundation Trust, University of Pavia (Italy), and Ludwig Maximilian University (Germany).
Public and patient involvement was central to the research, with advisory panels and youth groups—many with lived experience—providing crucial feedback on both ethical and practical aspects of the tool’s implementation.
A pilot phase is currently underway, testing the tool in real-world clinical settings and developing training materials for healthcare professionals through co-designed interventions. If successful, this tool could greatly reduce the number of people with unrecognised 'At Risk Mental states' and improve outcomes for thousands across the region.
The findings are published in the peer-reviewed journal Frontiers in Psychiatry, in the article titled Local adaptation and validation of a transdiagnostic risk calculator for first episode psychosis using mental health patient records.
