Background: Finding a suitably HLA-matched unrelated donor (URD) is essential for successful hematopoietic stem cell transplant (HSCT). Existing search productivity measures do not - or only partially - address real-world donor availability which can greatly affect a patient's transplant timeline.
Objective: This study aims to improve upon existing scores to provide a more accurate assessment of unrelated donor search productivity.
Study design: Search Summary Score (SSS) is a novel metric that integrates NMDP HapLogicSM HLA match probabilities with individual Donor Readiness Scores (DRS), a machine-learning based estimate of donor availability. SSS provides a probabilistic, patient-specific summary of expected search outcomes across multiple match levels.
Results: SSS was validated using 18 222 URD searches from 2021 to 2024, achieving superior predictive accuracy compared to NMDP Search Prognosis and 8/8 URD Search Prognosis (v1.0 and v2.0). For predicting 8/8 matches, SSS demonstrated an AUC of 0.962 and F1 score of 0.96. SSS utility is illustrated in examples of distinguishing clinically meaningful scenarios within the same Search Prognosis category.
Conclusions: SSS offers an interpretable, scalable, and globally applicable tool to assess matched donor availability at the initiation of search. By integrating both HLA compatibility and real-world donor readiness, SSS enhances clinical decision-making, promotes equity in donor access, and supports earlier planning for efficient selection of alternative graft sources.
Keywords: Donor availability; Hematopoietic stem cell transplantation (HSCT); Predictive modeling in transplantation; Search Productivity Measure; Search Summary Score (SSS); Unrelated donor (URD) search.
Copyright © 2025. Published by Elsevier Inc.