Enhancing Healthcare Decisions: Predicting Hospital Stay with Explainable ML

Efficient bed control in hospitals minimizes prices and enhances patient effects.This observation offers a predictive framework for Intensive Care Unit (ICU) period of life (LOS) at admission, leveraging digital fitness statistics (EHR) facts.Using the hospital live dataset from the Kaggle repository, diverse tool gaining knowledge of algorithms-which includes Logistic Regression, Random Forest, Multilayer Perceptron (MLP), Gradient Boosting, XGBoost, and a complicated CatBoost extension-had been evaluated primarily based on AUC, accuracy, precision, don't forget, and F1-rating.While XGBoost performed the very best accuracy among traditional models, the extended CatBoost set of policies outperformed all, reaching an outstanding 98.25% accuracy.CatBoost's potential to deal with specific information and optimize education capabilities through iterative selection timber appreciably progressed predictive universal performance.To make sure transparency and interpretability, Explainable AI (XAI) techniques, which includes SHAP, were hired to research feature contributions, imparting important insights into the factors influencing ICU stay length.This have a observe highlights the capacity of mixing affected individual EHR data with superior, interpretable system analysing models to decorate ICU stay predictions, ultimately supporting higher useful resource allocation and selectionmaking in hospitals.