<p>Effective medical waste management is critical for advancing healthcare sustainability, safeguarding the environment, and controlling operational costs. This study applies a combination of machine learning (ML) and explainable artificial intelligence (XAI) to predict both the volume and cost of medical waste in six private hospitals located in the Western Mediterranean region of Turkey. The dataset includes monthly records of waste categories—infectious, pathological, and cutting and piercing tool waste—along with associated cost components, spanning January 2020 to July 2025. Using these data, multiple ML algorithms were evaluated to forecast trends through 2031, with model performance assessed via Mean Absolute Error (MAE), Mean Squared Error (MSE), and the coefficient of determination (R²). Among all models tested, the Stacking Regressor achieved the highest predictive accuracy for both waste amounts and costs. Feature importance analyses consistently highlighted infectious waste metrics, particularly their cost and volume, as the most influential predictors. To enhance interpretability, SHAP and LIME were employed, offering both global and local explanations of model outputs and identifying operational factors with the greatest impact on waste generation and expenditure. By combining robust predictive performance with transparent model insights, this work addresses limitations of prior studies that relied on opaque or purely statistical models. The proposed framework offers a scalable, interpretable, and actionable tool to support data-driven decision-making, regulatory compliance, and sustainability-focused waste management in healthcare institutions.</p>