Prenatal Prediction of Umbilical Cord Blood CD34+ Adequacy and Identifying Influential Factors via An Ensemble Machine Learning and TOPSIS Ranking: A Retrospective Study
نویسندگان
1 Department of Industrial Engineering, Faculty of Engineering, Kharazmi University, Tehran, Iran
2 Department of Industrial Engineering, Faculty of Engineering, Kharazmi University, Tehran, Iran
3 Department of Industrial Engineering, Faculty of Engineering, Kharazmi University, Tehran, Iran
4 Department of Industrial Engineering, Faculty of Engineering, Kharazmi University, Tehran, Iran
5 Department of Stem Cells and Developmental Biology, Cell Science Research Center, Royan Institute for Stem Cell Biology and Technology, ACECR, Tehran, Iran
doi
10.22074/cellj.2025.2064741.1878چکیده
Objective: Prenatal prediction of CD34+ adequacy supports cord-blood banking by reducing expenses on low-yieldunits and reserving capacity for clinically promising grafts. We have developed and evaluated a prenatal machinelearning model that predicts whether an umbilical cord blood (UCB) unit will meet a clinically supported adequacythreshold (≥1.5×105 CD34+ cells/kg recipient) before collection for single-unit grafts.Materials and Methods: In this retrospective study, we analysed 126,406 records from the Royan Stem Cell TechnologyCompany (RSCT; Tehran, Iran), which included routinely available maternal, neonatal, and family-history variables. A pipelineof imputation (IterativeImputer numeric; SimpleImputer+OrdinalEncoder categorical), feature selection (Extra Trees), andhyperparameter tuning using Bayesian optimisation with model training/evaluation was performed within cross-validationfolds. Decision Tree (DT), K-Nearest Neighbours (KNN), Random Forest (RF), Support Vector Machine (SVM), and MultilayerPerceptron (MLP) classifiers were tuned via Bayesian optimisation. Models were ranked by the Technique for Order Preferenceby Similarity to Ideal Solution (TOPSIS). Majority voting (MV) ensembles were constructed from the top-k models. Modelinterpretability used SHapley Additive exPlanations (SHAP).Results: The MV (top-4: RF, KNN, DT, MLP) ensemble achieved an area under the receiver operating characteristiccurve (ROC-AUC)=0.808 and an area under the precision-recall curve (PR-AUC)=0.744 on the held-out test set, withan accuracy=0.757, precision=0.726, recall/sensitivity=0.804, F1=0.762, specificity=0.716, and Brier score=0.181.SHAP highlighted history of hepatitis C, birth place, hyperthyroidism, history of anaemia, oral fungus, and rheumatismamong the most influential features.Conclusion: Prenatal prediction of UCB CD34+ adequacy using an interpretable MV ensemble is feasible and accurateto support pre-collection triage and can potentially improve banking efficiency. The resultant model offers a non-invasivetool to enhance the efficiency of cord blood banking by prioritising units with higher transplantation potential.
