Heart failure & cardiomyopathies
Prediction of 90 Day Readmission in Heart Failure with Preserved Ejection Fraction by Interpretable Machine Learning
Authors' conclusion
Our model could help medical agencies achieve the early identification of 90 day readmission risk in HFpEF patients and reveal risk factors that provide valuable insights for treatments.
Study context
- Population
- Patients with heart failure with preserved ejection fraction (HFpEF), EHR data 1 August 2020 to 1 August 2021103 (13.8%) readmitted
- Participants
- 746 patients
- Comparison
- XGBoost vs random forest, neural network and logistic regression
- Outcome
- Readmission
- Follow-up
- 90 days after discharge
Methodological detailsmachine learning, electronic health records, Feature extraction by univariate analysis and least absolute shrinkage and selection operator (LASSO)
- Machine learning
- Electronic health records
- Feature extraction by univariate analysis and LASSO
Performance
As reported in the abstract
Prediction of 90-day readmission
| Test or model | AUC | Precision-recall area under the curve (PR-AUC) | Sensitivity | Specificity | Balanced accuracy | Kolmogorov-Smirnov (KS) statistic |
|---|---|---|---|---|---|---|
| XGBoost | 0.896 | 0.868 | 0.817 | 0.837 | 0.827 | 0.694 at 0.468 |
XGBoost
AUC0.896
Precision-recall area under the curve (PR-AUC)0.868
Sensitivity0.817
Specificity0.837
Balanced accuracy0.827
Kolmogorov-Smirnov (KS) statistic0.694 at 0.468
Further findings
Top 12 risk features (SHAP)
- Activities of daily living (ADL), left atrial dimension (LAD), LVDD, shortness, nitrates, length of stay, nutritional risk, fall risk, accompanied by other symptoms, educational level, anticoagulants, edema
Original publication
Prediction of 90 Day Readmission in Heart Failure with Preserved Ejection Fraction by Interpretable Machine Learning. ESC Heart Fail. Published 21 August 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →