Risk Prediction Models for the Onset of Mental Health Disorders: A Systematic Review of Model Methods,Performance and Clinical Utility
1
Division of Psychiatry, University College London, London, United Kingdom
Popul. Med. 2026;8(Supplement Supplement 1):A2560
ABSTRACT
ABSTRACT:
Common mental disorders (CMD) and severe mental illness (SMI) constitute a substantial public health concern globally, with approximately 1 in 7 individuals worldwide living with a mental health condition.¹ Mental health among young adults has been continuously declining over recent decades.² Early intervention has been shown to provide improved mental health and social outcomes³ and to be more effective than treatment as usual by reducing symptom severity and hospital admissions.⁴˒⁵ Risk prediction can enhance early identification at a population level ⁶ through the development of accurate prediction models that screen for mental health risk. A systematic review was conducted to identify and evaluate externally validated prediction models for the onset of depression, anxiety, bipolar disorder, or psychosis in young adults (<30 years). MEDLINE, APA PsycINFO, and Embase were searched up to October 2025, with independent screening conducted by two reviewers (Cohen’s κ = 0.85). Internal and external validation studies were paired for each identified model. Studies using population-based cohorts or case-control samples were included. Model performance was assessed using area under the curve (AUC), and risk of bias was evaluated for each study using the Prediction model Risk Of Bias ASsessment Tool (PROBAST). Out of 591 studies yielded by the search, 14 unique prediction models were identified for anxiety (n = 1), depression (n = 3), bipolar (n = 1) or psychosis (n = 9). Model discrimination ranged from poor to excellent (AUC: 0.59 - 0.92). While most models demonstrated acceptable to very good discrimination in internal validation (AUC > 0.70), substantially fewer showed comparable performance in external validation. Overall risk of bias was low across studies. These findings highlight the potential of high-performing prediction models to support early identification of mental health risk, with implications for scalable population-level prevention strategies, equitable intervention planning, and public mental health policy.
Common mental disorders (CMD) and severe mental illness (SMI) constitute a substantial public health concern globally, with approximately 1 in 7 individuals worldwide living with a mental health condition.¹ Mental health among young adults has been continuously declining over recent decades.² Early intervention has been shown to provide improved mental health and social outcomes³ and to be more effective than treatment as usual by reducing symptom severity and hospital admissions.⁴˒⁵ Risk prediction can enhance early identification at a population level ⁶ through the development of accurate prediction models that screen for mental health risk. A systematic review was conducted to identify and evaluate externally validated prediction models for the onset of depression, anxiety, bipolar disorder, or psychosis in young adults (<30 years). MEDLINE, APA PsycINFO, and Embase were searched up to October 2025, with independent screening conducted by two reviewers (Cohen’s κ = 0.85). Internal and external validation studies were paired for each identified model. Studies using population-based cohorts or case-control samples were included. Model performance was assessed using area under the curve (AUC), and risk of bias was evaluated for each study using the Prediction model Risk Of Bias ASsessment Tool (PROBAST). Out of 591 studies yielded by the search, 14 unique prediction models were identified for anxiety (n = 1), depression (n = 3), bipolar (n = 1) or psychosis (n = 9). Model discrimination ranged from poor to excellent (AUC: 0.59 - 0.92). While most models demonstrated acceptable to very good discrimination in internal validation (AUC > 0.70), substantially fewer showed comparable performance in external validation. Overall risk of bias was low across studies. These findings highlight the potential of high-performing prediction models to support early identification of mental health risk, with implications for scalable population-level prevention strategies, equitable intervention planning, and public mental health policy.
| eISSN: | 2654-1459 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
