From Coordination to Personalization: A Trust-Aware Simulation Framework for AI-Driven Personalized Decision Support...

Background/Objectives: Efficient and personalized task allocation in hospital emergency departments (EDs) is critical for operational efficiency and patient-centered care. However, the complexity of staff coordination and the variability among patients and healthcare professionals pose significant challenges. This study proposes a simulation-based framework for modeling doctors and nurses as intelligent agents guided by computational trust mechanisms. The objective is to explore how trust-informed coordination can support AI-driven and personalized decision-making in ED management. Methods: The framework was implemented in Unity, a 3D graphics platform, where agents assess their competence and patient-specific needs before undertaking tasks and adaptively coordinate with colleagues. The simulation environment enables real-time observation of workflow dynamics, resource utilization, and patient outcomes. We examined three scenarios-Baseline, Replacement, and Training-reflecting alternative staff management strategies. Results: Trust-informed task allocation balanced patient safety and efficiency by adaptively responding to nurse performance and patient acuity levels. In the Baseline scenario, prioritizing safety reduced errors but increased patient delays compared to a FIFO policy. The Replacement scenario improved throughput and reduced delays, though at additional staffing costs. The training scenario fostered long-term skill development among low-performing nurses, despite short-term delays and risks, supporting sustainable and personalized capacity building in ED teams. Conclusions: The proposed framework demonstrates the potential of computational trust for personalized and evidence-based decision support in emergency medicine. By linking staff coordination with adaptive and AI-informed decision-making, hospital managers are provided with a tool to evaluate alternative staffing and treatment policies under controlled and repeatable conditions. This work thus contributes to the broader vision of precision and personalized medicine, where operational decisions dynamically adapt to both patient needs and staff capabilities.