Machine learning prediction in cardiovascular diseases: a meta-analysis
Problem
Accurately predicting cardiovascular diseases including coronary artery disease, heart failure, stroke, and cardiac arrhythmias.
Why it matters
Cardiovascular diseases are major causes of death, so reliable prediction supports early prevention and treatment.
What the AI did
A meta-analysis pooled 103 cohorts (3,377,318 individuals) to quantify how ML algorithms predict cardiovascular diseases.
Results
Coronary artery disease: boosting AUC 0.88 (95% CI 0.84-0.91), custom-built AUC 0.93 (0.85-0.97); stroke: SVM AUC 0.92 (0.81-0.97).
Impact
Systematic meta-analysis across 103 cohorts provides aggregated evidence of ML's predictive ability, though not clinical deployment.
Record
| Venue | Scientific Reports |
|---|---|
| Published | 2020 |
| Work type | review |
| Authors | Chayakrit Krittanawong, Hafeez Ul Hassan Virk, Sripal Bangalore, Zhen Wang, Kipp W. Johnson, Rachel Pinotti, HongJu Zhang, Scott L. Kaplin et al. |
| Citations | 514 (OpenAlex, at import time) |
| References | 38 |
| Model or method | boosting, support vector machine, custom-built algorithms |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41598-020-72685-1 |
| Open access | gold |
| Keywords | machine learning, cardiovascular disease, meta-analysis, AUC, prediction |
Importance rating
major — millions of lives, an industry or a field
Abstract pools 103 cohorts with 3,377,318 individuals for ML prediction of major cardiovascular diseases (CAD, heart failure, stroke, arrhythmias), indicating large-scale impact.