Machine Learning-Based Prediction of Coronary Care Unit Readmission: A Multi-Hospital Validation Study
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Readmission to the Coronary Care Unit (CCU) has significant implications for patient outcomes and healthcare expenditure, emphasizing the urgency to accurately identify patients at high readmission risk. This study aims to construct and externally validate a predictive model for CCU readmission using machine learning (ML) algorithms across multiple hospitals. Patient information, including demographics, medical history, and laboratory test results were collected from electronic health record system and contributed to a total of 40 features. Three ML models, Logistic Regression, Random Forest, and Gradient Boosting were employed to estimate the readmission risk. The gradient boosting model was selected demonstrated superior performance with an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.887 in the internal validation set. Further external validation in hold-out test set and three other medical centers upheld the model’s robustness with consistent high AUCs, ranging from 0.852 to 0.879. The results endorse the integration of ML algorithms in healthcare to enhance patient risk stratification, potentially optimizing clinical interventions and diminishing the burden of CCU readmissions.
Key learning points
What is already known:
Readmission to the CCU has significant implications for both patient outcomes and healthcare costs.
Accurately distinguishing patients at high or low risk for CCU readmission is essential for clinicians to allocate resources effectively
What this study adds:
A predictive model for CCU readmission was constructed using machine learning algorithms trained from one medical center and validated externally in three major medical centers.
Among the ML models evaluated, the Gradient Boosting model showed the highest performance with an AUC of 0.879 in hold-out test set, and its robustness was further confirmed in external validation across three medical centers with an AUC range from 0.848-0.863.
By using different cut-off thresholds to prioritize the model’s sensitivity or specificity, clinicians can distinguish between high-risk and low-risk patients, enabling them to determine the appropriate level of monitoring and treatment planning for those at high risk.
