XGBoost machine learning algorithm for predicting unplanned readmission in elderly patients with coronary heart...

Area of Science:

• Cardiology
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Limited research exists on 1-year unplanned readmissions for elderly coronary heart disease (CHD) patients, with most studies focusing on 30-day readmissions.
• Extreme Gradient Boosting (XGBoost) models show promise for predicting patient outcomes due to their performance and interpretability.

Purpose of the Study:

• To develop and validate an XGBoost model for predicting 1-year unplanned readmissions in elderly Chinese patients diagnosed with CHD.
• To identify significant risk factors associated with 1-year unplanned readmissions in this patient cohort.

Main Methods:

• Retrospective collection of clinical data from elderly CHD patients.
• Feature selection using the stepwise forward method.
• Model performance evaluation using Area Under the Receiver Operating Characteristic Curve (AUROC) and Area Under the Precision-Recall Curve (AUPRC).
• Feature importance assessment using SHapley Additive exPlanations (SHAP).

Main Results:

• The study included 2137 elderly CHD patients.
• The XGBoost model achieved an AUROC of 0.704 and an AUPRC of 0.392.
• SHAP analysis identified length of stay (LOS), age-adjusted Charlson comorbidity index (ACCI), monocyte count, blood glucose, and red blood cell (RBC) count as key predictors.

Conclusions:

• The XGBoost model effectively predicts 1-year unplanned readmissions in elderly CHD patients.
• The model aids in identifying crucial risk factors for readmission, enabling targeted interventions.