Pre-existing and machine learning-based models for cardiovascular risk prediction
Problem
Cardiovascular risk prediction models needed assessment and improvement using large, complex healthcare data for primary prevention.
Why it matters
Accurate risk prediction guides prevention of cardiovascular disease, a leading cause of death.
What the AI did
Machine learning prediction algorithms were developed and benchmarked against pre-existing models on a large Korean adult cohort.
Results
Neural network C-statistic 0.751 vs Pooled Cohort Equation 0.738 (n=222,998), significantly higher, with improved calibration
Impact
Peer-reviewed cohort study showing a modest but significant improvement over the standard PCE model.
Record
| Venue | Scientific Reports |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Sang-Yeong Cho, Sun–Hwa Kim, Si‐Hyuck Kang, Si-Hyuck Kang, Kyong Joon Lee, Dongjun Choi, Seungjin Kang, Seungjin Kang et al. |
| Citations | 85 (OpenAlex, at import time) |
| References | 45 |
| Model or method | neural network (compared with logistic regression, treebag, random forest, adaboost) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41598-021-88257-w |
| Open access | gold |
| Keywords | cardiovascular risk, machine learning, C-statistic, prediction, cohort study |
Importance rating
major — millions of lives, an industry or a field
Primary prevention of cardiovascular disease guided a cohort of 222,998 adults; ML modestly improved prediction over standard models.