Innovative Applications of Unsupervised Learning in Uncertainty-Aware Pharmaceutical Supply Chain Planning

The significance of resiliency, reliability, and equity in the pharmaceutical supply chain is often overlooked but becomes evident in the wake of disastrous events. Disruptive incidents underscore the critical importance of these concepts, necessitating the development of innovative frameworks to effectively address the challenges that emerge in their aftermath. This paper introduces a framework specifically designed to address the issues arising from disruptions within the pharmaceutical supply chain. A novel mixed-integer nonlinear programming (MINLP) model is proposed to formulate the pharmaceutical supply chain that encompasses the distribution of both cold and non-cold pharmaceuticals and vaccines. The abundance of diverse pharmaceuticals and vaccines, each with its distinct characteristics, presents a formidable planning obstacle. A noteworthy contribution of this study lies in innovatively applying AI-driven methodologies to pharmaceutical supply chain, employing five pioneering unsupervised learning algorithms for improved inventory management and control. The model's uncertainty is effectively addressed through an innovative joint chance constraint (JCC) formulation. By employing JCC, the model ensures a high level of reliability in satisfying uncertain patient demands. The MINLP formulation with JCCs presents significant computational complexities and intractability. To alleviate these issues, state-of-the-art reformulation algorithms are provided to transform the model into its equivalent mixed-integer linear programming form. The results indicate the efficiency of the equivalent reformulation techniques and illustrate the capabilities of the model to alleviate the resiliency, reliability, and equity concerns.