Predictive Modeling of Patient Health Outcomes Using Electronic Health Records and Advanced Machine Learning Algorithms

Abstract

Electronic Health Records (EHRs) provide a rich source of real-time patient data, offering unprecedented opportunities to develop predictive models for health outcomes. In this study, we explore the application of advanced machine learning (ML) algorithms to analyze and predict patient health trajectories. We compare a suite of models logistic regression, random forests, gradient boosting, and deep neural networks on a real-world EHR dataset to identify key clinical predictors and forecast patient outcomes such as hospital readmissions, length of stay, and mortality. Our results indicate that ensemble and deep learning methods outperform traditional approaches, offering enhanced predictive accuracy and model interpretability through SHAP (SHapley Additive exPlanations) values. The findings demonstrate the potential of ML-driven decision support systems in improving patient care, resource allocation, and proactive healthcare management.