Designing an efficient supply chain network for public cord blood bank and quality prediction to improve performance: a case study in Iran

Cord blood (CB), a vital source of hematopoietic stem cells, is increasingly used to treat various hematologic diseases. However, the limited number of CB cells remains a problematic factor restricting their use. Managing public cord blood banks and transplantation centers is a significant challenge for all countries. This study introduces a multi-objective mixed-integer linear programming model for designing the public cord blood supply chain. The model is developed to optimize performance by reducing costs and unmet demands, as well as by improving accessibility. Various machine learning algorithms have also been incorporated to evaluate and predict CB quality. A solution approach based on Lagrangian relaxation and the augmented ε-constraint method has been developed to overcome problem complexity and solve large-scale instances. The proposed model's practicality is validated through a case study in Iran. The results demonstrate the optimal location and allocation of network facilities. Comparative analysis reveals that the proposed supply chain model fulfills 37% more demand than the current system. Among the four quality predictive models analyzed, the K-nearest neighbor (KNN) method stood out, delivering an impressive 98.39% accuracy and a 90.85% AUC (Area Under the Curve) score. Implementing this method could potentially reduce 3% of the total costs of the supply chain. Sensitivity analyses are also provided to assess the trade-offs between objective functions and to offer valuable insights for decision-makers.