ZeBRA: Zero-burden EHR-based prediction of Chronic Kidney Disease onset and progression

ZeBRA (Zero-Burden Risk Assessment) uses routinely collected electronic health record (EHR) and administrative claims data to predict chronic kidney disease (CKD) and progression to severe stage of CKD. It analyzes longitudinal diagnosis, medication, and procedure histories without requiring additional laboratory tests, questionnaires, or manual data entry for scoring. The CKD work addresses two complementary needs: identifying patients at risk of a CKD diagnosis within one year, and identifying patients with early-stage CKD who are likely to progress. Potential applications include prioritizing nephrology review, targeting monitoring and preventive care, and identifying patients for clinical-trial prescreening. Evidence to date: • One-year CKD prediction: retrospective national-data AUCs of 0.818–0.830, with AUCs of 0.776–0.817 in an independent Chicago health-system cohort. • CKD progression: retrospective national-data AUCs of 0.801–0.819 for progression from stages 1–3 to advanced CKD over one month to two years, with AUCs of 0.745–0.794 in the Chicago cohort. • The national progression analysis included approximately 18,300 cases and 152,000 controls, aged 35–75. Reported AUC ranges reflect male and female models. The broader ZeBRA platform also has experience moving from research into health-system implementation. Its pulmonary model has been independently reconstructed and validated by University of Kentucky Medical Center’s data science team using Epic-derived data and is undergoing implementation for prospective silent evaluation. This provides a practical foundation for adapting the CKD models to clinical environments; the kidney application remains at the retrospective validation stage. Because ZeBRA uses structured records already available in many healthcare systems, it supports transfer through local data mapping, validation, and recalibration. Planned prospective evaluation would assess subgroup performance, clinically manageable review thresholds, referral and monitoring workflows, and downstream outcomes. We seek pharmaceutical, biotechnology, health-system, and digital health partners for external validation, prospective pilots, nephrology decision support integration, and clinical-trial enrichment studies. Opportunities include sponsored research, co-development, and university-coordinated licensing.