Comparison of risk models for mortality and cardiovascular events between machine learning and conventional logistic regression analysis
Problem
Predicting all-cause mortality and major cardiovascular events (heart failure, coronary syndrome, stroke, hemorrhage) in patients, and comparing machine learning with conventional logistic regression risk models.
Why it matters
Accurate risk stratification guides treatment decisions for patients with serious cardiovascular conditions.
What the AI did
Non-linear machine learning models identified 118 risk factors and their weightings for five clinical outcomes in a cohort of 15,933 patients from the Shinken Database.
Results
AUCs of 0.900 (mortality), 0.912 (heart failure), 0.879 (acute coronary syndrome), 0.758 (ischemic stroke), 0.753 (intracranial hemorrhage); comparison figures vs logistic regression not stated in abstract.
Impact
ML models achieved high predictive discrimination for mortality and cardiovascular events in this single cohort.
Record
| Venue | PLoS ONE |
|---|---|
| Published | 2019 |
| Work type | article |
| Authors | Shinya S. Suzuki, Takeshi Yamashita, Tsuyoshi Sakama, Takuto Arita, Naoharu Yagi, Takayuki Otsuka, Hiroaki Semba, Hiroto Kano et al. |
| Citations | 69 (OpenAlex, at import time) |
| References | 35 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1371/journal.pone.0221911 |
| Open access | gold |
| Keywords | machine learning, cardiovascular risk, mortality prediction, logistic regression, AUC |
Importance rating
major — millions of lives, an industry or a field
Predicting mortality and major cardiovascular events guides treatment for serious conditions; disease burden scale not stated, only cohort size.