Integrated Multi-Task Predictive Analytics fo...

Integrated Multi-Task Predictive Analytics for Electronic Health Records

Author Name : S. Akash, Dr. R. Sri Devi

DOI: https://doi.org/10.56025/IJARESM.2025.130225124

ABSTRACT Electronic Health Records (EHR) offer a vast repository of patient data that holds tremendous potential for improving clinical decision-making and operational efficiency. In this study, we propose a multi-task machine learning framework that simultaneously addresses the prediction of patient readmission risk, classification of diseases based on admission diagnoses, and forecasting of discharge destinations. The framework begins with an extensive data preprocessing stage—handling missing values, encoding categorical variables, and engineering new features such as Body Mass Index (BMI)—followed by feature scaling. Three distinct predictive models are deployed: a Random Forest Classifier for readmission risk, a Decision Tree Classifier for disease classification, and an XGBoost Classifier for discharge destination prediction. Through rigorous evaluation using metrics like precision, recall, and F1-score, our integrated approach demonstrates promising performance, offering insights that could significantly enhance patient care and hospital resource management.