Blood donation projections using hierarchical time series forecasting: the case of Zimbabwe’s national blood bank.

Publication date: Apr 01, 2024

The discrepancy between blood supply and demand requires accurate forecasts of the blood supply at any blood bank. Accurate blood donation forecasting gives blood managers empirical evidence in blood inventory management. The study aims to model and predict blood donations in Zimbabwe using hierarchical time series. The modelling technique allows one to identify, say, a declining donor category, and in that way, the method offers feasible and targeted solutions for blood managers to work on. The monthly blood donation data covering the period 2007 to 2018, collected from the National Blood Service Zimbabwe (NBSZ) was used. The data was disaggregated by gender and blood groups types within each gender category. The model validation involved utilising actual blood donation data from 2019 and 2020. The model’s performance was evaluated through the Mean Absolute Percentage Error (MAPE), uncovering expected and notable discrepancies during the Covid-19 pandemic period only. Blood group O had the highest monthly yield mean of 1507. 85 and 1230. 03 blood units for male and female donors, respectively. The top-down forecasting proportions (TDFP) under ARIMA, with a MAPE value of 11. 30, was selected as the best approach and the model was then used to forecast future blood donations. The blood donation predictions for 2019 had a MAPE value of 14. 80, suggesting alignment with previous years’ donations. However, starting in April 2020, the Covid-19 pandemic disrupted blood collection, leading to a significant decrease in blood donation and hence a decrease in model accuracy. The gradual decrease in future blood donations exhibited by the predictions calls for blood authorities in Zimbabwe to develop interventions that encourage blood donor retention and regular donations. The impact of the Covid-19 pandemic distorted the blood donation patterns such that the developed model did not capture the significant drop in blood donations during the pandemic period. Other shocks such as, a surge in global pandemics and other disasters, will inevitably affect the blood donation system. Thus, forecasting future blood collections with a high degree of accuracy requires robust mathematical models which factor in, the impact of various shocks to the system, on short notice.

Open Access PDF

Concepts Keywords
Donations Blood donation
Mathematical Bottom-up
Monthly Hierarchical forecasting
Pandemic Optimal combinations
Zimbabwe Top-down

Semantics

Type Source Name
disease IDO blood
disease VO time
disease VO monthly
disease MESH Covid-19 pandemic
pathway REACTOME Reproduction
drug DRUGBANK Coenzyme M
drug DRUGBANK Tretamine
disease MESH uncertainty
disease VO effective
drug DRUGBANK Hexadecanal
disease VO Ply
disease VO Optaflu
drug DRUGBANK Elm
drug DRUGBANK Esomeprazole
drug DRUGBANK Vildagliptin
disease IDO process
drug DRUGBANK L-Tyrosine
drug DRUGBANK Parathyroid hormone
drug DRUGBANK L-Valine
drug DRUGBANK Sodium Tetradecyl Sulfate
disease VO BT1
disease VO frequency
disease VO population
disease MESH emergencies
drug DRUGBANK Iron
disease IDO intervention
disease MESH shock
drug DRUGBANK Etoperidone
disease VO vein
disease VO Canada
disease VO Pal
disease VO organ
drug DRUGBANK Trestolone

Original Article

(Visited 1 times, 1 visits today)