Key result
Supervised machine-learning algorithms like logistic regression (c-index 0.921; 95% CI 0.920-0.922) showed higher discriminant validity for predicting incident MI than unsupervised techniques.
Why the study?
Incident and recurrent MI remain major global health issues, and efforts to improve risk prediction in population health studies are needed to help scale prevention strategies and management.
Can machine-learning algorithms accurately predict incident and recurrent myocardial infarction in a large US population?
Observational (n=4,300,000)
Yes
Can machine-learning algorithms accurately predict incident and recurrent myocardial infarction in a large US population?
Supervised machine-learning algorithms, particularly logistic regression and neural networks, demonstrate excellent discriminant validity for predicting incident and recurrent myocardial infarction in a large, diverse population.
Answers are grounded in this paper and cite their sources.
This large-scale observational analysis should not yet change practice for MI risk prediction; leaves open whether.
BACKGROUND: To date, incident and recurrent MI remains a major health issue worldwide, and efforts to improve risk prediction in population health studies are needed. This may help the scalability of prevention strategies and management in terms of healthcare cost savings and improved quality of care. METHODS: We studied a large-scale population of 4.3 million US patients from different socio-economic and geographical areas from three health plans (Commercial, Medicare, Medicaid). Individuals had medical/pharmacy benefits for at least 30 months (2 years for comorbid history and followed up for 6 months or more for clinical outcomes). Machine-learning (ML) algorithms included supervised (logistic regression, neural network) and unsupervised (decision tree, gradient boosting) methodologies. Model discriminant validity, calibration and clinical utility were performed separately on allocated test sample (1/3 of original data). RESULTS: In the absence of MI in comorbid history, the overall incidence rates were 0.442 cases/100 person-years and in the presence of MI history, 0.652. ML algorithms showed that supervised formulations had incrementally higher discriminant validity than unsupervised techniques (e.g., for incident MI outcome in the absence of MI in comorbid history: logistic regression "LR" - c index 0.921, 95%CI 0.920-0.922; neural network "NN" - c index 0.914, 95%CI 0.913-0.915; gradient boosting "GB" - c index 0.902, 95%CI 0.900-0.904; decision tree "DT" - c index 0.500, 95%CI 0.495-0.505). Calibration and clinical utility showed good to excellent results. CONCLUSION: ML algorithms can substantially improve the prediction of incident and recurrent MI particularly in terms of the non-linear formulation. This approach may help with improved risk prediction, allowing implementation of cardiovascular prevention strategies across diversified sub-populations with different clusters of complexity.
No takes yet. Share an insight, caveat, or question.
Lip et al. (2022) conducted an observational in Incident and recurrent myocardial infarction (n=4,300,000). Supervised machine-learning algorithms (logistic regression, neural network) vs. Unsupervised machine-learning algorithms (decision tree, gradient boosting) was evaluated on Discriminant validity (c-index) for incident MI in the absence of MI history (c-index 0.921, 95% CI 0.920-0.922). Supervised machine-learning algorithms like logistic regression (c-index 0.921; 95% CI 0.920-0.922) showed higher discriminant validity for predicting incident MI than unsupervised techniques.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: