Development of an artificial intelligence based model for predicting the euploidy of blastocysts in PGT-A treatments

Abstract

The euploidy of embryos is unpredictable before transfer in in vitro fertilisation (IVF) treatments without pre-implantation genetic testing (PGT). Previous studies have suggested that morphokinetic characteristics using an artificial intelligence (AI)-based model in the time-lapse monitoring (TLM) system were correlated with the outcomes of frozen embryo transfer (FET), but the predictive effectiveness of the model for euploidy remains to be perfected. In this study, we combined morphokinetic characteristics, morphological characteristics of blastocysts, and clinical parameters of patients to build a model to predict the euploidy of blastocysts and live births in PGT for aneuploidy treatments. The model was effective in predicting euploidy (AUC = 0.879) but was ineffective in predicting live birth after FET. These results provide a potential method for the selection of embryos for IVF treatments with non-PGT.