TIMO: Thrombocytes Inventory Management Tool for Blood Banks

Blood product management is a problem of general human interest with a number of concerns and complications. By far the most dominant concern is the risk for infections and contamination. One other concern is the production and inventory management of blood platelets (thrombocytes). These platelets are essential for wound repair (coagulation) when a blood vessel is damaged or when there is a deficiency such as for hematology patients.

Here, however, there are two conflicting aspects. On the one hand, despite planned elective surgeries, the demand is highly ‘uncertain' and roughly over 50% (at week basis) remains to be unpredictable. Clearly, as lives may be at risk, shortages should not take place at all or at least should be kept to a minimum (say less than 1%).

On the other hand, as supply is voluntary, blood is to be considered as highly precious. In addition, costs are involved with the production step. As such, any spill by outdating will be unethical and highly undesirable. General outdating figures in the Western world (USA and Western Europe) range from roughly 20% down to 4%. This outdating is to be reduced if not minimized.

In practical blood management environments (Blood Banks and hospitals) simple replenishment (or order-up-to) rules are generally used. For each day of the week these rules prescribe a production up to some fixed replenishment level.

The values of these replenishment levels still remain to be determined and in practice are merely based upon experience. So far there seemed to be no formal support for the computation of these values and to which extent such simple rules are near to optimal.

SDP-Simulation approach

To this end, in [1] a mathematical formulation has been proposed as based upon the technique of stochastic dynamic programming (SDP). This technique computes an optimal production level for each possible state x which keeps track of the ages of the platelets available (see figure 1 below).

However, for practical purposes this technique has two major complications:

1. The size of the state space and the number of computations, in the order of 10⁵⁰, is prohibitively large for realistic computation.
2. The ‘optimal policy' is far more complex than just a simple order-up-to rule.

A simulation based approach has therefore also been developed (see the technical paper [1] which integrates the mathematical computations with simulation to provide:

• A nearly optimal policy (say within 2% optimality)
• A simple order-up-to replenishment rule

Results

A first case-study for the Dutch North-East Blood Bank showed a substantial improvement over a one year period of the outdating figure well below 1% while the shortage also remained below 1% (see [2]).

A second case-study for the Dutch South-East Blood Bank showed similar figures. In addition, though, as a third performance aspect, for quality reasons also the age, at which platelets were issued, was considered. By optimizing different scenarios by simulation, strategies were obtained, by which the mean issuing age was brought down by over 1 day (see [3]), while keeping shortages and outdating below 1%. This reduction could be regarded as a considerable quality improvement as platelets deteriorate (clutter) with age. Particular patient groups (such as for oncology and hematology) require the best possible quality of (i.e. youngest) platelets.

User-friendly software tool (TIMO)

Motivated by these positive results a user-friendly software program has been developed for the Dutch South-East Blood Bank. Within this program, the numerical procedures for the mathematical (SDP) formulations and the simulation procedures, as originally developed by R. Haijema from the University of Amsterdam, were integrated and automated in one user-friendly software tool in cooperation between the University of Amsterdam (Nikky Kortbeek) and INCONTROL Simulation Solutions (Fred Jansma).

This tool, called TIMO (Thrombocytes Inventory Management Optimizer), has meanwhile been implemented and will be tested with practical data for a half year period. It has a general set-up for extension to other Blood Banks. Also other applications of this SDP-Simulation approach for perishable products can be thought of such as within the retail industry. As such it provides another illustration of the fruitful combination and potential of techniques for optimization as based upon simulation.

Acknowledgement. The cooperation and the involvement of the Dutch Blood Bank Sanquin, region South East, under the responsibility of the director Wim de Kort has been highly appreciated.