Psychosis, characterised by chronic symptoms often emerging in youth, imposes a substantial burden on individuals and healthcare systems. While early detection and intervention can mitigate this burden, there is limited evidence on the cost-effectiveness of such approaches. To address this lack of evidence, this study protocol outlines the health economic implications of an artificial intelligence (AI)-based intervention, the Computer-Assisted Risk-Evaluation (CARE), designed to prevent psychosis. The intervention uses AI technologies to enhance the diagnosis and treatment quality for individuals at high risk of psychosis.

The health economic evaluation has been designed alongside a 12-month multicentre randomised controlled trial comparing CARE with treatment as usual from both payer and societal perspectives. An implementation cost analysis will complement the evaluation, and long-term consequences beyond the trial will be explored descriptively. Based on a literature review, an initial economic logic model will guide subsequent analyses by depicting CARE’s programme theory.

The cost-effectiveness assessment will include averted cases of manifest psychosis and quality-adjusted life-years using the EuroQol 5-Dimensions 3-Level instrument. Other effectiveness outcomes will also be incorporated into a cost–consequence analysis. Cost-effectiveness acceptability curves reflecting statistical uncertainty will be constructed, incorporating various payer and societal willingness-to-pay values. The implementation cost analysis will follow a mixed-methods approach to capture facility-specific costs.

A dark logic model, emphasising negative outcomes, will be developed to investigate long-term consequences. Further, the initial economic logic model will be refined using trial data and expert interviews. This comprehensive approach aims to provide decision-makers not only with evidence on the cost-effectiveness of CARE, but also with a broader understanding of the implications of the intervention.

The study has received ethical approval and plans to disseminate its findings through publication in a peer-reviewed journal and conference presentations.

NCT05813080.