SIMULATION OPTIMISATION OF ULTRASONOGRAPHY RESOURCE SCHEDULING WITH MACHINE LEARNING

Abstract

This study proposes a decision-support framework to optimize radiologist staffing in the ultrasonography department. Patient arrival rates were forecast using Light Gradient Boosting Machine (LightGBM) with feature expansion, achieving 99.99 % accuracy over one-month period. A discrete-event simulation model was subsequently used to determine the number of radiologists required to meet target waiting times. Based on the simulation results, hourly radiologist requirements were identified, and an optimized schedule was generated. By integrating machine learning, simulation, and scheduling, this framework supports data-driven planning and can be applied to other healthcare services and facilities facing demand uncertainty.