Strategic Selection of Resilient Pharmaceutical Replenishment Policies: A Human-in-the-Loop (HITL) Z-Number SWOT-VIKOR Framework

Abstract

In pharmaceutical logistics, supply chain replenishment policies must balance financial efficiency with systemic resilience. Traditional policy selection relies on cost-centric multi-criteria decision-making (MCDM) frameworks that are susceptible to qualitative biases and hierarchical mathematical distortion. This study proposes a Human-in-the-Loop (HITL) Z-Number SWOT-VIKOR framework, synergizing human strategic governance with AI computational scalability. Human experts establish macro-strategic SWOT weights via the F-CIMAS method, while Large Language Models (LLMs) determine micro-operational weights anchored in CRITIC data variance. Evaluating nine Vendor-Managed Inventory (VMI) policies under stochastic demand, the extended VIKOR algorithm proposed a compromise set of four policies, governed by the hybrid SMT-SSP policy ( ) which strictly minimized individual regret ( ). Methodologically, this study contributes the novel "Strategic Modulation" mechanism, successfully resolving the internal "erasure effect" of classical hierarchical MCDM. By dynamically scaling baseline weights rather than applying strict categorical normalization, this mechanism preserved intrinsic data variance, preventing a statistically significant rank distortion ( ) in mid-tier policies. Furthermore, external mathematical benchmarking against TOPSIS and SAW algorithms proved that VIKOR's non-compensatory regret-minimization is strictly necessary for pharmaceutical logistics. The framework provides supply chain managers with a mathematically protected, bias-resistant blueprint for strategic decision-making in high-stakes healthcare environments.