Transformer-based models for predicting cardiovascular risk in Chinese adults: development and validation.
AI trained on millions of Chinese patients predicts heart disease risk better than established scores, tailored to populations most affected.
The China-AIHeart transformer model was trained on 156,790 participants from the China Cardiometabolic Disease and Cancer Cohort and externally validated in two independent Chinese cohorts, demonstrating superior 10-year cardiovascular risk prediction compared with both traditional Cox models using the same predictors and established risk scores including China-PAR and PREVENT-ASCVD. The sex-specific architecture captures temporal and interaction patterns that Cox regression cannot, providing a practical AI-enhanced risk stratification tool for the world's largest high-CVD-burden population.
What the study was
- Study design
- development_and_external_validation_cohort
- Category
- cardiovascular_metabolic
- Maturity
- Validated
- Journal
- Eur Heart J
Why it surfaced
European Heart Journal publication; robust 156K derivation cohort plus two independent external validations; direct head-to-head vs guideline risk scores with significant NRI improvement; addresses CVD prediction gap in Chinese/Asian populations; dual-topic overlap AI/ML + cardiovascular.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.