A Stochastic modeling framework for ICU resource allocation during health crises

The unprecedented surge in demand for intensive care services during health crises such as the COVID-19 pandemic has revealed critical limitations in existing ICU resource allocation models, which often fail to adapt to uncertain and dynamic conditions. This study aims to develop and evaluate a stochastic modeling framework to optimize ICU resource allocation under crisis scenarios, accounting for probabilistic patient arrivals, fluctuating treatment durations, and constrained multi-resource environments. The framework integrates discrete-event simulation (DES), queueing theory (specifically M/M/c/K models), and stochastic optimization to simulate real-time ICU operations and support decision-making. A Monte Carlo simulation was conducted over a 24-hour period involving 100 replications, where key parameters included a patient arrival rate of 4 patients/hour, 5 ICU beds, and a service time distribution with an average of 6 hours. The results indicate a high blocking probability of 84.3%, ICU bed utilization of 94%, ventilator utilization of 90%, an average patient waiting time of 2.4 hours, and a delay-sensitive mortality rate of 8%. The expected system cost, incorporating waiting time, mortality, and resource inefficiency penalties, totaled 190 units. These findings demonstrate the model’s capability to reveal critical system bottlenecks and support adaptive, ethically grounded allocation policies. The proposed framework provides practical implications for hospital administrators and policymakers by offering a dynamic, evidence-based decision-support tool to improve ICU efficiency and patient outcomes during emergencies.