Morphological assessment is still the gold standard for embryo selection. Artificial Intelligence (AI) has been developed for embryo selection. However, AI is still a complementary method of supporting humans. This study aims to investigate a combination of the Early Embryo Viability Assessment (EEVA) AI and blastocyst morphological assessment as a predictor of pregnancy outcomes of single vitrified-warmed blastocyst transfer. The retrospective cohort study was conducted in a single center from 2020 to 2023 and included 511 single vitrified-warmed blastocyst transfer cycles. Blastocyst transfer quality was based on morphology. Embryos on Day 3 were evaluated using the EEVA system. The correlation between the EEVA system alone, blastocyst morphological assessment alone or a combination of the EEVA system and blastocyst morphological assessment, and pregnancy outcomes was qualified by GEEs. Comparison of 3 methods to evaluate the results of predicting pregnancy outcomes using the area under the curve (AUC): performed on the prediction probability scale of the model rule. The GEE model using the EEVA system showed a negative association between higher EEVA scores and the likelihood of achieving pregnancy outcomes. Embryos with the highest EEVA score (EEVA 5) have substantially lower odds of achieving successful implantation and ongoing pregnancy compared with those with the lowest score (EEVA 1). The OR of Score 5 vs 1 was 0.282 (95% CI 0.125–0.636, p < 0.001) for implantation and 0.228 (95% CI 0.092-0.563, p < 0.001) for ongoing pregnancy. The AUC of the GEE model using the EEVA system for implantation and ongoing pregnancy potential was 0.651 and 0.655, respectively. The AUC of the GEE model using the blastocyst morphological assessment for implantation and ongoing pregnancy potential was 0.703 and 0.700, respectively. The AUC of the GEE model combining both systems for implantation and ongoing pregnancy potential was 0.730 and 0.726. The differences were statistically significant (p < 0.001). The EEVA system can predict the success rates, especially when combining EEVA with blastocyst morphological assessment in blastocyst selection for transfer.
Vu et al. (Tue,) studied this question.