Scalable risk stratification of undiagnosed heart failure using routine health data and its association with imaging phenotypes and outcomes.
A simple computer model using existing medical records can identify undiagnosed heart failure across millions of patients in different countries.
FIND-HF demonstrates that a simple logistic regression model using routine EHR variables can scalably identify undiagnosed heart failure across diverse health systems globally, with AUC performance matching more complex approaches. The external validation in >18M person-dataset pool across four countries is one of the most geographically broad HF screening validations published, and the link between model-derived risk scores and CMR imaging phenotypes provides mechanistic credibility for clinical adoption.
What the study was
- Study design
- retrospective_multicohort_external_validation
- Population
- General primary care populations across multiple countries
- Sample size
- >18 million (multi-cohort)
- Category
- Diagnostics
- Maturity
- Validated
- Journal
- Scientific Reports
Why it surfaced
Scalable population-level HF risk stratification from EHR is directly applicable to health system deployment and cardiometabolic pipeline; validates across NHS, US Epic Cosmos, and Asian health systems; performance parity with complex models suggests near-term NHS/payer adoption potential.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.