This study introduces a predictive framework for emergency department operations by integrating advanced machine learning methodologies. Utilizing logistic regression, the model forecasts patient admission dispositions from routine clinical data to improve resource allocation and patient flow. Additionally, the framework employs dimensionality reduction and clustering techniques—UMAP, Gaussian Mixture Models, and PCA—to reveal latent patient subgroups, offering deeper insights into data distributions and enabling targeted clinical interventions.
Predicting ED Admission Disposition and Patient Clustering for Resource Optimization
Kasra Afzali 