Choodoung M.Promwong C.Wongba K.Choodoung A.Kerdpin U.Thichanpiang P.Plabplueng C.Fichou Y.Nuchnoi P.Mahidol University2025-09-132025-09-132025-09-01Journal of Applied Laboratory Medicine Vol.10 No.5 (2025) , 1200-1214https://repository.li.mahidol.ac.th/handle/123456789/112044Background The D-elution (DEL) phenotype is serologically mislabeled as Rh-negative because of the very low amount of D antigen on red blood cells. The adsorption-elution test and genotyping are recommended tests for confirmation. However, turnaround time and the availability of instruments, reagents, and budget, as well as technical issues are challenging factors of DEL identification in laboratory practice and patient safety. Methods To develop a screening predictive algorithm for DEL and Rh-negative, the serological tests of RhCcEe antigen and adsorption-elution tests were computed using a machine learning model. Results The machine learning algorithm computed the data based on RhCcEe antigen with or without a DEL confirmative serological test like the adsorption-elution test. The predictive accuracy gave >90% for RhD-negative identification in a Thai blood donor dataset. To screen for RhD-negative, we provided the web application named RhDnostics at https://rnp-project-1.streamlit.app/. Conclusion Our machine learning algorithm could be used as a predictive tool for RhD-negative screening in the laboratory with no confirmative serological test or RHD molecular testing available.Chemical EngineeringChemistryBiochemistry, Genetics and Molecular BiologyMedicineArticleSCOPUS10.1093/jalm/jfaf0742-s2.0-10501521245424757241