Healthcare systems are increasingly complex. Evaluating hospital performance is now a strategic necessity to ensure accountability, efficiency, and better patient outcomes. Traditional expert-based methods, while valuable, often suffer from subjectivity, high time costs, and a lack of scalability. This study presents a novel hybrid framework. It integrates Bayesian Best-Worst Method (BBWM), Multi Attributive Ideal-Real Comparative Analysis (MAIRCA), and Boosted Decision Tree Regression (BDTR) to create a transparent, data-driven, and predictive hospital evaluation model. Initially, eight key performance indicators were selected and weighted using BBWM to reflect expert preferences under uncertainty. MAIRCA was then employed to calculate performance scores by comparing hospitals’ actual performance to an ideal reference point. These scores were used to train the BDTR model, allowing for accurate performance prediction based solely on input indicators. The proposed method was applied to data from 250 Iranian hospitals. Results showed that the BDTR model achieved high predictive accuracy, with an R2 of 0.697 and a Mean Absolute Error (MAE) of just 0.090. Feature importance analysis highlighted that financial efficiency, operational costs, and bed utilization were the strongest predictors of hospital performance. Compared to traditional Decision Tree (DT) models, the boosting technique improved prediction accuracy by over 10.8%, and it also outperformed Random Forest (RF) and Extreme Gradient Boosting (XGBoost) by 3.5% and 1.6%, respectively. The proposed model supports scalable performance monitoring and benchmarking, and the case study demonstrates its practical viability for health policymakers and hospital administrators seeking to implement evidence-based evaluation systems.