Analyzing 30-Day Readmission Rate for Heart Failure Using Different Predictive Models.

Area of Science:

• Health Services Research
• Medical Informatics
• Cardiology

Background:

• Hospital readmissions are a key performance metric for the Centers for Medicare and Medicaid Services.
• Heart Failure (HF) is a leading cause of costly hospitalizations and readmissions.
• Accurate risk stratification at discharge can enable timely post-discharge interventions.

Purpose of the Study:

• To compare the predictive performance of two risk stratification models for 30-day HF readmissions.
• To evaluate logistic regression and random forest models using clinical predictors from electronic health records.

Main Methods:

• Utilized electronic health records data from 1037 Heart Failure patients.
• Developed and compared logistic regression with bagging and random forest models.
• Assessed model performance using C-Statistics.

Main Results:

• Logistic regression with bagging achieved a higher predictive accuracy (C-Statistics: 0.65) compared to random forest (C-Statistics: 0.61).
• The models were built using 48 clinical predictors.

Conclusions:

• Logistic regression with bagging demonstrates superior performance for predicting HF readmissions.
• This approach can improve identification of HF patients at high risk for readmission.