Improving supply chain resilience and delivery performance in the medical device industry : a ripple effect simulation and ai-based decision support approach

Abstract

Medical device supply chains are particularly vulnerable to disruptions due to their globalized structure, regulatory constraints, and reliance on international sourcing. This dissertation investigates disruption propagation within the Wing-to-Wing (W2W) supply chain of GE HealthCare Algeria, an import-dependent, zero-inventory, make-to-order system. A three-pillar methodology is developed. The first pillar applies a Fuzzy ANP–Fuzzy TOPSIS approach to prioritize previously identified disruptions according to their impact, likelihood, propagation effect, and controllability, while establishing an order-centric stochastic representation of the supply chain to formalize disruption propagation. The second pillar develops a discrete-event simulation model in AnyLogic to quantify disruption impacts on delivery performance. The third pillar introduces SC Copilot, an AI-powered decision-support system integrating large language models with simulation outputs for scenario analysis and managerial decision-making. Results reveal a strong asymmetry in disruption behavior, where missing-item disruptions generate substantially greater delays than factory delays by triggering a complete second import cycle. The study contributes a methodological framework for disruption prioritization, stochastic modeling, and simulation-based analysis in zero-inventory supply chains, while demonstrating the potential of agentic AI to enhance resilience and decision- making in the medical device sector.