Key result
Machine learning algorithms modestly improve ASCVD risk prediction vs PCE in healthy adults.
Why the study?
Predicting cardiovascular disease risk is key to primary prevention, and machine learning offers potential for analyzing complex healthcare data to improve risk prediction.
Do machine learning-based algorithms improve cardiovascular risk prediction compared to pre-existing models in statin-naïve healthy adults without cardiovascular disease?
Population
222,998 Korean adults aged 40–79 years, naïve to lipid-lowering therapy, with no history of cardiovascular disease
Comparison
Machine learning-based prediction algorithms vs pre-existing cardiovascular risk prediction models
Authors
Loading...
ML modestly boosts discrimination over PCE in Koreans; leaves open whether neural networks should supplant traditional equations in practice.
Cohort (n=222,998)
No
Do machine learning-based algorithms improve cardiovascular risk prediction compared to pre-existing models in statin-naïve healthy adults without cardiovascular disease?
Effect estimate: C-statistic 0.751 (95% CI 0.740-0.761)
p-value: p=<0.001
Machine learning-based algorithms, particularly neural networks, provide modest but significant improvements in discrimination and calibration for predicting 5-year cardiovascular risk compared to traditional models like the Pooled Cohort Equations.
Cho et al. (2021) conducted a cohort in Cardiovascular disease risk prediction (n=222,998). Machine learning-based algorithms vs. Pooled cohort equation (PCE) and other contemporary models was evaluated on First hard atherosclerotic cardiovascular disease event (C-statistic 0.751, 95% CI 0.740-0.761, p=<0.001). Machine learning-based algorithms improved cardiovascular risk prediction with a C-statistic of 0.751 compared to Pooled Cohort Equations showing 0.738 in statin-naïve healthy Korean adults without cardiovascular disease.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: