Resilient Coordination of Test Sampling and Requirement Supply in the Lab Test Supply Chain: Soft Worst-Case Distributionally Robust Optimization

Experiences from the global pandemic highlight the need to restructure blood sampling for CRP tests to adapt to varying levels of epidemic severity and operational risks. Additionally, since the tests rely on specific requirements, it’s crucial to establish a coordinated framework for both networks to ensure effective operations. This study develops a model comprising three subproblems to address the outlined problem. The first subproblem involves a blood sample collection model, focusing on regional coverage and route optimization under operational risks. The second subproblem assesses suppliers for the SC network of lab-test requirements, funneling top candidates to the third subproblem. The supply chain for CRP test requirements is established in the third subproblem, incorporating optimal test and sample numbers from Subproblem 1 and top suppliers from Subproblem 2. The proposed model is designed with a resilient structure to address pandemic challenges by incorporating redundancies in the supply process, minimizing unmet demand, and planning for capacity expansion. Infection rate uncertainty significantly impacts the CRP sampling model. To address this, the study introduces a distributionally robust optimization model with a soft worst-case approach, minimizing worst-case violations while reducing conservatism. In the second subproblem, suppliers are evaluated using a novel hybrid MCDM model. This model considers essential factors such as information capability, service quality, access speed, supplier capacity, supplier robustness, and economic advantage. To demonstrate the applicability of the proposed study, the model is applied to a case study in Tehran. The findings reveal the optimal locations, capacities, and types of facilities, their optimal paths within both networks, and their optimal interrelations with suppliers to meet demand requirements effectively. In addressing infection uncertainty, the proposed robust model demonstrated a clear reduction in conservatism regarding expectation violations. This allows for better control of risk aversion while maintaining a flexible, soft approach to modeling.