Established machine learning models to predict readmission for elderly patients with ischemic heart disease
Xuewu Song1, Feng Xian2, Changyu Zhu1
1Department of Pharmacy, Sichuan Provincial People's Hospital, University of Electronic Science and Technology of China, Chengdu, China.
Kardiologia Polska
|July 8, 2024
Summary
Machine learning models can identify elderly ischemic heart disease patients at high risk for 30-day or 1-year readmission using discharge data. Key clinical features help predict readmission risk.
Area of Science:
- Cardiology
- Medical Informatics
- Machine Learning
Background:
- Predicting readmission risk in elderly ischemic heart disease (IHD) patients is crucial but understudied.
- Clinical features influencing 30-day or 1-year readmissions require further investigation.
Purpose of the Study:
- Develop and validate machine learning models for predicting 30-day and 1-year readmissions in elderly IHD patients.
- Identify key clinical features for readmission risk prediction using routinely collected hospital discharge data.
Main Methods:
- Employed eight machine learning algorithms to construct prediction models.
- Assessed model performance using Area Under the Receiver Operating Characteristic Curve (AUROC) and Area Under the Precision-Recall Curve (AUPRC).
- Utilized SHapley Additive exPlanations (SHAP) for feature contribution analysis.
Main Results:
- The Categorical Boosting (CB) model demonstrated superior predictive performance for both 30-day (AUROC 0.72) and 1-year (AUROC 0.66) readmissions.
- Identified common important features including age-adjusted Charlson comorbidity index, brain natriuretic peptide, heart failure, and cholesterol.
- Other significant predictors included free thyroxine, thymidine kinase 1, osmotic pressure, and red blood cell distribution width.
Conclusions:
- Machine learning models effectively identify elderly IHD patients at high risk for readmission.
- Routinely collected discharge data can be leveraged for accurate readmission risk stratification.
