A digital twin framework integrated with a mixed proactive-reactive model for human milk supply chain planning

Abstract

Human milk is critical for infant development, particularly for premature infants or those whose mothers face supply challenges. Existing literature on the human milk supply chain (HMSC) typically relies on static approaches that overlook the inherent uncertainties and dynamic nature of the network. Consequently, this paper proposes a novel framework that integrates simulation and optimization techniques within a digital HMSC layer to dynamically compute optimal recipe levels required to maintain high service standards in neonatal intensive care units (NICUs). The simulation models the process of milk deposit collection at human milk banks (HMBs) by incorporating various stochastic factors to reflect real-world complexities. Subsequently, the mathematical model is implemented in two phases within the physical HMSC. In the proactive phase, production recipes are generated based on the capacities and requirements of HMBs, and delivered to NICUs. In the reactive phase, a clustering approach among NICUs is developed, coupled with lateral transshipments, to prevent shortages and reduce wastage. By applying a rolling horizon approach, analysis based on data from Tehran province demonstrates that the proposed framework outperforms alternative methods. Compared to a digital twin that excludes the reactive phase, our framework achieves superior performance by integrating lateral transshipments. In conclusion, our proposed framework has the potential to mitigate supply disruptions, reduce unmet demand, and maintain high service levels within the network.