Optimizing Inter-Regional Blood Transfers in a Stock-and-Flow Simulation Model of the U.S. Blood Supply

• Activity Number: 612
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Risk Analysis
• Author(s): Arianna Simonetti*+ and Hussein Ezzeldin and Richard Forshee
• Companies: FDA/CBER and FDA/CBER and FDA
• Keywords: Blood Supply ; Stock and Flow Model ; ABO groups ; Rh type ; Neural Network Optimization

Abstract

An adequate and resilient blood supply is critical for public health. Decision makers need tools to assess how to minimize blood supply disruptions during a time sensitive crisis. We developed an inter-regional stochastic stock-and-flow simulation model to evaluate potential shocks to the US blood supply, either in aggregate or by ABO/Rh type. The model consists of a simplified network of five regions each represented by two stocks: a blood collector and a hospital. A previous algorithm of blood transfers, which accommodated the needs of additional blood from simulated regions with the surplus supply of other regions, failed to achieve an overall optimal network performance. Sub-optimal transfers among regions were resolved by using advanced optimization methods. We propose a Neural Network Heuristic approach (NNH) to optimize the allotment of blood among regions, and increase network transfers efficiency. The NNH optimizes the network global performance by using a multivariate weighted cost function that combines factors such as stocks inventory levels; projected supply and demand; blood expiration and shortages. Simulation results on a mass casualty event will be presented.