Risk Alert System to Facilitate Universal Suicide Risk Screening in Emergency Care: Protocol for a Prediction Model and Software Development Study Using a Co-Design Approach

Background Suicide is a leading cause of preventable mortality and a major contributor to years of life lost. Many people who later die by suicide present to emergency departments in the months before death, often for reasons not explicitly related to self-harm and without receiving a risk assessment. Universal suicide risk screening in emergency care can improve detection but is difficult to implement because of time constraints, workflow disruption, and limited capacity to respond to increased identification. Objective The Clinical Risk Alert System for Suicide Risk Screening in Emergency Settings (CARES) project aims to lower the practical threshold for universal screening by developing a risk alert system software prototype for use among patients presenting to emergency care, prompting a brief, validated suicide risk screening and clinician-led assessment. Methods CARES will develop machine learning–based prediction models for subsequent nonlethal intentional self-harm and suicide within 1, 6, and 12 months after discharge from an index emergency department visit, using linked, population-based electronic registries from Catalonia (Spain), including electronic health records, mortality data, administrative sociodemographic variables, and a specific self-harm register. Index visits will include emergency department contacts in individuals aged 6 years or older without self-harm–related chief complaints, with predictors defined from information recorded in the prior 12 months. Models will be trained with approaches addressing rare outcomes and unequal sampling probabilities, and evaluated using an independent test set and temporal validation. The risk alert system software prototype will be designed as a web-based application with a backend hosting trained models and a frontend that allows data entry and displays risk as descriptive text and visualizations in absolute and relative terms compared with same-age and same-sex peers. Implementation research will use a co-design process with a user advisory group (UAG; clinicians, people with lived experience, caregivers, and stakeholder organizations), guided by the Medical Research Council framework for complex interventions, to define alert thresholds, response pathways, usability requirements, and training needs. Results The CARES project was funded in March 2023. The project uses linked electronic registry data from Catalonia covering 2014-2019, including 2,982,736 eligible emergency department index visits from 603,098 patients. The first UAG meeting was held in December 2024. As of May 2026, the project is developing the initial prediction models and software backend, while iteratively refining the frontend through UAG meetings. Data analysis is ongoing, and the main results are expected to be published in spring 2027. Conclusions CARES will deliver an experimental proof-of-concept risk alert system designed to support future targeted suicide risk assessment within emergency workflows while keeping clinical decision-making with clinicians and patients. If feasible and acceptable, this approach could improve detection of otherwise unrecognized risk and inform future real-time integration and evaluation within routine emergency care. International Registered Report Identifier (IRRID) DERR1-10.2196/100498