O-297 Prediction of pregnancy outcomes of single vitrified-warmed blastocyst transfer using combination of an automatic classification algorithm applied on cleavage stage embryos and blastocyst morphology

Abstract Study question Does combination of embryo assessment algorithm on cleavage-stage based on the automatic annotation of morphokinetic and blastocyst morphology in embryo selection improve implantation outcomes? Summary answer Combination of embryo assessment algorithm on cleavage-stage based on the automatic morphokinetic annotation and blastocyst morphological evaluation was showed significantly predictive for implantation outcomes. What is known already Conventional morphological assessment is still the gold standard for embryo selection. Timelapse monitoring of embryo development may represent a superior way to culture and select embryos in vitro. Nowadays, time-lapse culture systems are becoming more popular, many algorithms have been developed for embryo selection based on embryo morphokinetics, providing additional information for morphological assessment. However, manual annotations of developmental events and application of algorithms can be time-consuming and subjective processes. Automation of morphokinetic annotation can be a potential approach that reduces subjectivity in embryo selection. Study design, size, duration The observational, retrospective study was conducted in a single centre between 2020-2023 and included 511 single vitrified-warmed blastocyst transfer cycles. Blastocyst selected for transfer on Day 5,6 based on conventional morphological evaluation. Embryos were evaluated on Day 3 with a score from 1 (best) to 5 (worst) (from highest to lowest developmental potential) by the automatic embryo assessment algorithm. The correlation of the algorithms scoring and pregnancy outcomes was qualified by generalized estimating equations (GEEs). Participants/materials, setting, methods Embryos were cultured in timelapse GERI incubator and classified on day 3 by automatic embryo assessment algorithm depending on four parameters: P2 (t3-t2), P3 (t4-t3), oocyte age, and number of cells. The correlation of the algorithms scoring and pregnancy outcomes was qualified by generalized estimating equations (GEEs). Three GEE model using embryo assessment algorithm, conventional morphological assessment, and a combination of both classification system as the predictor were compared. Main results and the role of chance The implantation rate was higher with lower the scores generated by the embryo assessment algorithm. A GEE model showed the negative association between higher embryo score and lower odds of implantation. The OR of Score 3; 4; 5 vs 1 were 0.335; 0.259; 0.245 (95% CI 0.194-0.580; 0.138-0.485; 0.111-0.544, P = 0.000), respectively, for implantation. The AUC of the model using embryo assessment algorithm for implantation potential = 0.641. The AUC of the model combining both embryo assessment algorithm and conventional morphological assessment for implantation potential = 0.733. The differences were statistically significant (z-statistic = 3.896, p = 0.0001). Limitations, reasons for caution The sample size was enough to confirm the ability of the model for embryo selection. However, the retrospective nature of this study may be a limitation. Wider implications of the findings Using combination of timelapse technology with automated embryo assessment and conventional morphological assessment can predict the success rates of assisted reproduction cycles. To our knowledge, this is the first study using embryo dataset from single vitrified-warmed blastocyst transfer cycles with this embryo assessment algorithm. Trial registration number not applicable