LIDIA — Costa Rica's Predictive AI Flags Undiagnosed Diabetes in Public Clinic Records

Costa Rica's Caja Costarricense de Seguro Social (CCSS), the public agency that runs the country's universal health system, has since 2023 been piloting LIDIA, an internal program of predictive AI models built on top of EDUS, its nationwide Single Digital Health Record. LIDIA currently covers four condition areas: type 2 diabetes, lung health, acute coronary syndrome and breast cancer.
In its diabetes module, LIDIA's algorithm cross-references clinical history and risk factors across more than a million EDUS records to identify patients likely to have or develop type 2 diabetes before a formal diagnosis is made. A pilot at the Clorito Picado clinic applied the tool to the clinic's patient population; among the group the algorithm flagged as high-risk, clinical fieldwork found that 130 patients already had diabetes that had not yet been formally diagnosed. Clinic director Dr Carlos Solano Salas said that, had the tool been applied earlier, the disease could have been prevented or caught sooner in some of those cases.
The independent AI Observatory Costa Rica lists LIDIA as a "verified adoption/pilot" as of August 2026. Following the diabetes pilot, CCSS has said it intends to extend the same predictive approach to its other three modules — lung disease, acute coronary syndrome and breast cancer — and has paired the rollout with staff training across medical, IT and administrative departments; a related clinical decision-support tool, AIDA, was announced for integration with EDUS in November 2025. No peer-reviewed accuracy, sensitivity or false-positive rate for the diabetes model has been published, and neither CCSS nor the AI Observatory has disclosed a bias audit, so accuracy figures circulating in some secondary coverage have not been independently verified against a published methodology.

Read the full analysis: https://thecostaricanews.com/artificial-intelligence-has-arrived-in-the-costa-rican-healthcare-system-pilot-plan-at-the-clorito-picado-clinic-identifies-130-patients-with-diabetes/