Explainable haemoglobin deferral predictions using machine learning models: Interpretation and consequences for the blood supply

Abstract

Background and objectives: Accurate predictions of haemoglobin (Hb) deferral for whole-blood donors could aid blood banks in reducing deferral rates and increasing efficiency and donor motivation. Complex models are needed to make accurate predictions, but predictions must also be explainable. Before the implementation of a prediction model, its impact on the blood supply should be estimated to avoid shortages.

Materials and methods: Donation visits between October 2017 and December 2021 were selected from Sanquin's database system. The following variables were available for each visit: donor sex, age, donation start time, month, number of donations in the last 24 months, most recent ferritin level, days since last ferritin measurement, Hb at nth previous visit (n between 1 and 5), days since the nth previous visit. Outcome Hb deferral has two classes: deferred and not deferred. Support vector machines were used as prediction models, and SHapley Additive exPlanations values were used to quantify the contribution of each variable to the model predictions. Performance was assessed using precision and recall. The potential impact on blood supply was estimated by predicting deferral at earlier or later donation dates.

Results: We present a model that predicts Hb deferral in an explainable way. If used in practice, 64% of non-deferred donors would be invited on or before their original donation date, while 80% of deferred donors would be invited later.

Conclusion: By using this model to invite donors, the number of blood bank visits would increase by 15%, while deferral rates would decrease by 60% (currently 3% for women and 1% for men).

Keywords: blood donation testing; donor health; haemoglobin measurement.

FIGURE 1

Performance metrics for all models (left: Women, right: Men). (a): Precision of class non‐deferral; the proportion of successful donations among all predicted non‐deferrals. The complement of the precision is the deferral rate, should the model be used to guide invitations. (b): Recall of class non‐deferral; the proportion of successful donations that are predicted correctly. The complement of the recall is the proportion of missed donations, should the model be used to guide invitations. Note that the y‐axes in are zoomed in to highlight the differences between various models

FIGURE 2

SHAP summary plots for predictions made by SVM‐5, on 100 random donations from the test set. Each point represents one single observed donation. The location on the x‐axis indicates the contribution of the predictor variable on the prediction (positive value: Indicative class non‐deferral, negative: Indicative of class deferral) while the colour of the point indicates the relative value of the feature in that observation. The features on the y‐axis are ordered by their relative importance, measured as the mean absolute SHAP value

FIGURE 3

Cumulative distribution of the timing of donor invitations on basis of first predicted positive Hb level relative to the original donation date