a novel hybrid algorithm for designing a sustainable supply chain of car-t therapy in a multi-objective mode considering disease relapse

This study investigates the sustainable supply chain network design problem in the healthcare sector, where patients with cancer are treated using car-t cell therapy. to better reflect real-world conditions, the possibility of disease relapse is incorporated into the problem formulation. the problem is modeled as a multi-objective mixed-integer programming (mip) problem, aiming to minimize total costs, reduce environmental impacts, and maximize social satisfaction and accessibility. given the np-hard nature of the problem, a novel hybrid metaheuristic algorithm is developed to solve large-scale instances. the proposed algorithm is a structured integration of three evolutionary methods: non-dominated sorting genetic algorithm iv (nsga-iv) for preserving diversity and pareto front coverage, s-metric selection evolutionary multi-objective algorithm (sms-emoa) for enhancing precision and hypervolume expansion, and the epsilon-dominance evolutionary multi-objective algorithm (ε-moea) for rapid initial convergence. these three were selected as their combined strengths ensure a balanced trade-off between exploration, convergence speed, and final accuracy, which cannot be achieved by any of them individually. the proposed hybrid algorithm employs a two-stage selection mechanism, an adaptive mutation strategy, and a dynamic external archive to generate high-quality solutions across the pareto front. numerical experiments across different problem scales confirm its superiority, yielding on average 30% more non-dominated solutions, a 3% reduction in costs, and a 2 to 3% decrease in environmental impacts compared to single algorithms. the findings demonstrate this hybrid approach potential to enhance both strategic and operational decision-making in resilient healthcare delivery networks.