Key result
MLM-PREVENT machine learning model improves 10-year ASCVD risk calibration over standard AHA-PREVENT.
Why the study?
Does a machine learning-adapted AHA-PREVENT model improve calibration and risk reclassification for 10-year ASCVD events compared to the original AHA-PREVENT model in a local population without prior ASCVD?
Population
95,326 patients without prior atherosclerotic cardiovascular disease from a New England-based electronic…
Comparison
MLM-PREVENT vs Original AHA-PREVENT model
Design
Cohort
Follow-up
10 years (event period 2007-2016)
Authors
Loading...
An interpretable machine learning approach successfully recalibrated the AHA-PREVENT model for a local population, improving calibration and reclassifying 11.5% of patients at the 7.5% risk threshold while preserving original risk associations.
Cohort (n=95,326)
Does a machine learning-adapted AHA-PREVENT model improve calibration and risk reclassification for 10-year ASCVD events compared to the original AHA-PREVENT model in a local population without prior ASCVD?
p-value: P = .53 vs P < .001
An interpretable machine learning approach successfully recalibrated the AHA-PREVENT model for a local population, improving calibration and reclassifying 11.5% of patients at the 7.5% risk threshold while preserving original risk associations.
Zinzuwadia et al. (2024) conducted a cohort in Without prior atherosclerotic cardiovascular disease (ASCVD) (n=95,326). MLM-PREVENT model vs. AHA-PREVENT model was evaluated on Calibration for ASCVD events (nonfatal MI, coronary artery disease, ischemic stroke, or cardiovascular death) (p=P = .53 vs P < .001). The MLM-PREVENT machine learning model improved calibration for 10-year ASCVD risk compared to the AHA-PREVENT model in the overall cohort (Hosmer-Lemeshow P=0.53 vs P<0.001).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: