Machine learning-based prediction of coronary care unit readmission: A multihospital validation study

Area of Science:

• Cardiology
• Medical Informatics
• Machine Learning

Background:

• Coronary care unit (CCU) readmissions significantly impact patient outcomes and healthcare costs.
• Accurate identification of patients at high risk for CCU readmission is essential for timely intervention.

Purpose of the Study:

• To develop and externally validate a predictive model for CCU readmission.
• To utilize machine learning (ML) algorithms for enhanced patient risk stratification.

Main Methods:

• Collected patient data from electronic health records, encompassing demographics, medical history, and lab results (40 features).
• Evaluated five ML models: logistic regression, random forest, support vector machine, gradient boosting, and multilayer perceptron.
• Selected the gradient boosting model for its superior performance.

Main Results:

• The gradient boosting model achieved an area under the receiver operating characteristic curve (AUC) of 0.887 in internal validation.
• External validation across multiple centers confirmed the model's robustness with AUCs ranging from 0.852 to 0.879.
• The model demonstrated consistent high performance in predicting CCU readmissions.

Conclusions:

• Machine learning algorithms can effectively enhance patient risk stratification in healthcare settings.
• The developed predictive model shows promise for optimizing clinical interventions and reducing CCU readmission rates.
• Integration of ML in clinical practice can lead to improved patient management and resource allocation.