Description:
Clinical laboratories play a key role in the delivery of healthcare services, as the result of clinical tests affect 70 percent of medical decisions. Therefore, to provide high quality services, small labs send some of their samples to a well-equipped central lab. The objective of this paper is to minimize the total logistics costs when collecting blood samples from geographically dispersed locations. Because blood samples are perishable, the collection process must be completed within two hours. As these labs do not own the vehicle fleet, a third-party logistics (TPL) service provider is used in the collection process. We define this problem as the robust multi-capacitated time-constraint open vehicle routing problem with hub location and demand prediction (RMCTCOVRP-HLDP). The three main contributions of our work are: first, the development of a new open vehicle routing problem (OVRP) model in which sequences of open routes and their intersections (creating hub nodes) are allowed, second, the use of Bertsimas and Sim's robust approach to deal with travel cost uncertainty, And third, using predicted values for labs' demands obtained from training several machine learning algorithms on historical data in Python as a way to deal with demands uncertainty. The mathematical model is solved using CPLEX solver in GAMS and the optimal solution is obtained. Moreover, the sensitivity analysis on three essential parameters proves the validity of our model. Computational results are also compared with that of the equivalent classical OVRP model and show that our model always performs better