JMIR Publications

ABSTRACT

Background:

Platelets are a valuable and perishable blood product. Managing platelet inventory is a demanding task due to short shelf lives and high variation in daily platelet use patterns. Predicting platelet demand is a promising step toward avoiding obsolescence and shortages and ensuring optimal care.

Objective:

The aim of this study is to forecast platelet demand for a given hospital using both a statistical model and deep neural network. In addition, we aim to calculate possible reduction in waste and shortage of platelets under use of the predictions in a retrospective simulation of the platelet inventory.

Methods:

Predictions of daily platelet demand are made by a LASSO (Least Absolute Shrinkage and Selection Operator) model and by a recurrent neural network (RNN) with Long-short Term Memory (LSTM), respectively. Both models use the same set of 81 clinical features. Predictions are passed to a simulation of the blood inventory to calculate possible reduction of waste and shortage as compared to historical data.

Results:

From January 01, 2008 to December 31, 2018 the average waste and shortage rates for platelets were 10.1 % and 6.5 %, respectively. In simulations of platelet inventory waste could be lowered to 4.9 % with the LASSO and 5.0 % with the RNN while shortages were 2.1 % and 1.7 % with LASSO and RNN, respectively. Daily predictions of platelet for the next 2 and 4 days had mean average percentage errors of 25.5 % and 18.1 % with the LASSO and 26.3 % and 19.2 % with the RNN, respectively.

Conclusions:

Both models allow for predictions of platelet demand with similar and sufficient accuracy to significantly reduce waste and shortage in a retrospective simulation study. The possible improvements in platelet inventory management are roughly equivalent to 250,000 dollar per year.