Predicting the length of hospital stay (LOS) is critical for improving healthcare resource management and patient care. This study investigates the application of explainable machine learning techniques to forecast hospital stay duration using a dataset from Kaggle, comprising various patient and hospital-related features. The primary goal is to develop accurate predictive models and elucidate the underlying factors influencing hospital stay lengths. The study employs multiple machine learning algorithms, including Logistic Regression, Multi-Layer Perceptron (MLP), Random Forest, Gradient Boosting, and XGBoost. Each model's performance is evaluated using standard metrics such as accuracy, precision, recall, and F1-score. Additionally, explainability tools such as SHapley Additive exPlanations (SHAP) are utilized to interpret model predictions and identify the most significant predictors of LOS. The findings demonstrate that advanced machine learning models, particularly ensemble methods, achieve superior predictive accuracy. Moreover, the explainability analysis provides valuable insights into the critical factors influencing hospital stays, thereby enabling healthcare practitioners to make informed decisions and optimize hospital resource allocation. This research underscores the potential of integrating explainable machine learning into healthcare analytics to enhance operational efficiency and patient outcomes.

Keywords: Hospital Stay Length, Machine Learning, Logistic Regression, MLP, Random Forest, Gradient Boosting, XGBoost, Explainability, SHAP, Healthcare Analytics.