Key result
Machine learning models using clinical, laboratory, and ECG data predict inpatient PCI need with ~0.76 AUC.
Why the study?
The study was conducted to evaluate the feasibility of predicting PCI using clinical, laboratory, and electrocardiographic data available at different stages of hospitalization.
Can machine learning models using non-invasive clinical, laboratory, and ECG data predict the need for percutaneous coronary intervention in patients hospitalized with suspected coronary artery disease?
Population
137 patients with suspected coronary artery disease
Comparison
Basic vs clinical vs extended data models across three machine learning algorithms
Design
Retrospective single-center observational study
Authors
Loading...
Should not yet alter PCI strategies in diabetic patients; leaves open key questions for future randomized trials.
Observational (n=137)
No
Can machine learning models using non-invasive clinical, laboratory, and ECG data predict the need for percutaneous coronary intervention in patients hospitalized with suspected coronary artery disease?
Effect estimate: ROC-AUC 0.755
Machine learning models utilizing early non-invasive clinical, ECG, and glycemic data demonstrate moderate feasibility for predicting the need for PCI in patients with suspected coronary artery disease.
Alimbayeva et al. (2026) conducted an observational in suspected coronary artery disease (n=137). Machine learning models (logistic regression, random forest, gradient boosting) was evaluated on Percutaneous coronary intervention (PCI) performed during the current hospitalization (ROC-AUC 0.755). Machine learning models using clinical, laboratory, and ECG data predicted the need for percutaneous coronary intervention with moderate accuracy (best ROC-AUC 0.755).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: