Predicting Readmission Among High-Risk Discharged Patients Using a Machine Learning Model With Nursing Data:...

Area of Science:

• Health Informatics
• Machine Learning in Healthcare
• Clinical Decision Support

Background:

• Unplanned hospital readmissions increase healthcare costs and reduce care quality.
• Proactive discharge planning from admission is crucial for reducing readmission risks.
• Machine learning models offer enhanced predictive power for preemptive discharge care.

Purpose of the Study:

• To develop an early readmission prediction model using nursing data for high-risk patients.
• To compare the performance of various machine learning algorithms for readmission prediction.

Main Methods:

• Retrospective study of 12,977 patients with high-risk readmission diseases (Jan 2018-Jan 2020).
• Utilized demographic, clinical, and nursing data to build two prediction models: Model 1 (early data) and Model 2 (all data).
• Employed logistic regression, random forest, decision tree, XGBoost, CatBoost, and multiperceptron layer algorithms with 5-fold cross-validation and adaptive synthetic sampling.

Main Results:

• The CatBoost model in Model 2 achieved the highest predictive performance with an AUROC of 0.64.
• The Random Forest model in Model 1 (early prediction) had an AUROC of 0.62.
• Key predictors included BMI, systolic blood pressure, and age; nursing data variables were more prominent in the early prediction model.

Conclusions:

• Machine learning models incorporating nursing data provide essential support for clinical decisions in managing high-risk discharge patients.
• These models facilitate early interventions by offering comprehensive risk assessments.
• Integrating diverse nursing data can improve patient outcomes and reduce readmission rates.