Approximate Dynamic Programming Modeling for a Typical Blood Platelet Bank

In this paper, we have devised a real-time, workable model for solving the practical problems associated with responsible blood platelet inventory. These problems include how to eciently dispense, organize, store, and order platelets that become unusable after 6 days. Stochastic demand and supply, as well as deterministic lead times, are congured into their model. Any shortages or outdates of the 8 major blood types are penalized in this proposal, with revenue in conjunction with a further emphasis on the age of platelets will maximize ecency as well. A combined model of linear programing and approximate dynamic programming (ADP) was deployed while constructing their practical model. In an environment which emphasizes the need for real-time judgements, this new policy, as advanced here, will lead to substantial shortage, outdate, and cost reductions of these time-stamped platelets. The experimental application of this model produced shortages in the range of approximately 4.7% with outdates at 5.5%, a vast improvement over existing methods. An ADP approach was found to be practical during this research, and additionally the model showed much promise with respect to enhancing a reward system that simultaneously decreases shortages and the number of expired platelets.