PHYSICIAN ASSIGNMENT AND SCHEDULING UNDER LICENSURE CONSTRAINTS IN EMERGENCY TELEMEDICINE

Abstract

Telemedicine enables healthcare providers to deliver critical services across wide geographic areas, connecting patients in need with remote physicians in real time. In emergency teleneurology, stroke patients arriving at healthcare facilities are linked to off-site physicians through a centralized navigation system that assigns on-duty, credentialed physicians to time-sensitive cases within strict response windows. Failure to assign within the allowed time results in penalties, referred to as “blasts.” The first part of this study focuses on the real-time dispatching of remote on-duty physicians to emergency cases across multiple facilities. We develop and evaluate five dynamic physician-to-patient assignment policies using a discrete-event simulation framework grounded in real-world operational constraints, including stochastic case arrivals, licensure restrictions, and non-preemptive service logic, with the goals of (i) maximizing service levels and (ii) maintaining fairness in workload distribution. Performance metrics, namely, blast rate reduction and workload equity to assess and compare these policies. Among these, a novel predictive policy, Blast Prediction (PRED), leverages system state and demand distribution to anticipate potential overloads. Results show that system-aware, predictive assignment strategies can substantially improve blast rates and workload balance among physicians without requiring additional staff or infrastructure, demonstrating the value of adaptive dispatch logic in high-stakes telemedicine systems. With the understanding that the navigator’s decision-making is not only complicated by the physicians’ credentialing portfolios but also their heterogeneity in productivity, the second part of this study addresses the tactical scheduling of on-duty physicians by considering their productivity and facility-specific operational factors. Physician productivity, measured in patients per hour (PPH), varies with individual characteristics, such as experience and stress levels, and on-site conditions such as availability of scribes and registration protocols. To capture these dynamics, we propose a mixed-integer linear programming (MILP) model and a Hybrid Genetic Algorithm (HGA) to solve the Physician Roster Problem (PRP). The model incorporates credentialing portfolios, productivity indices, and hospital-level performance factors. Additionally, our model accounts for the dynamic, non-stationary nature of productivity, which varies by shift type (day or night). Using a comprehensive numeric analysis, we investigate the sensitivity of key parameters on physician schedules and the computational performance of our method. Moreover, we conduct a case study based on a real-life application by a telemedicine service provider operating in the United States and serving a variety of healthcare facilities across multiple states. Results demonstrate that productivity-aware rostering leads to substantial reductions in staffing shortages and improved service levels across facilities. Together, these two studies advance the operational design of emergency telemedicine systems by addressing decision-making challenges at both the real-time assignment and tactical scheduling levels. The integrated insights from dispatching and rostering optimization offer a cohesive framework for enhancing physician responsiveness, reducing operational costs, and improving patient outcomes in distributed telemedicine networks.