Sex-specific cardiovascular disease risk prediction using statistical learning and explainable artificial intelligence: the HUNT Study
Authors' conclusion
By employing statistical learning and incorporating sex-specific risk factors, we propose improved risk prediction models for CVD in males.
Study context
- Population
- Adults aged 40–79 years without prior cardiovascular disease (CVD), HUNT Study (Norway)
- Participants
- 31,946 adults
- Comparison
- XGBoost and logistic regression models vs NORRISK 2 and SCORE2
- Outcome
- Prediction of 10-year CVD risk (ROC-AUC and PR-AUC on the test set)
- Follow-up
- 10 years
Methodological detailsrandom training/test split, sex-specific models
- Random training/test split
- Sex-specific models
Performance
As reported in the abstract
CVD risk prediction, males
- XGBoost vs NORRISK 2 · males; PR-AUC increase 12%
- increase in ROC-AUC 0.013 unit
- XGBoost vs SCORE2 · males; PR-AUC increase 11%
- increase in ROC-AUC 0.012 unit
Further findings
Predictors
- All traditional NORRISK 2 and SCORE2 risk factors were important predictors for males, but not for females; potential new risk predictors identified
Age- and sex-specific intervention thresholds
- Improved sensitivity vs NORRISK 2-suggested thresholds
Original publication
Sex-specific cardiovascular disease risk prediction using statistical learning and explainable artificial intelligence: the HUNT Study. Eur J Prev Cardiol. Published 10 March 2025. doi:10.
Summarised from the published abstract; every number is checked against it