O-096 Effect of Day-3 embryo morphology on blastocyst quality and ploidy: a machine-learning model of 34781 embryos

Abstract Study question What is the predictive value of using Day-3 (D3) embryo morphology and clinical factors to identify good quality euploid blastocysts? Summary answer Using a machine-learning model, D3-cell count fragmentation helps identifying embryos most least likely develop into blastocysts. known already It well-known that pattern cleavage affects blastocyst ploidy. Less has been elucidated in relation fragmentation. Published studies often used outdated biopsy techniques (D3-biopsy) obsolete genetic technologies (FISH). Hence, association between ploidy status remains controversial. Existing hypothesis regulation event maintain homeostasis normalize constitution. Literature lacking investigating this newer such as Next Generating Sequencing (NGS). Moreover, model D3-embryo characteristics can be useful for triage laboratory resources. design, size, duration Retrospective observational single center study including 34781 from 5701 cycles, March 2017 2020. Embryos were evaluated degree cell number on D3, 68±1hrs post insemination. Female Age, AMH BMI annotated patient characteristics. All blastocysts available D5 or D6 with ≥ BL3CC (n = 18361) subjected trophectoderm (TE) Preimplantation-Genetic-Testing aneuploidies (PGT-A) analysis Next-Generation-Sequencing rates recorded. Participants/materials, setting, methods Degree was classified follows: A( < =10%), B(11-25%), C(26-35%), D(>35%). Data stratified according category: <6cells, 6-10 cells, >10 cells compacted stage. Morphology markers (cell-count rate), along (age, anti-Mullerian hormone body-mass index) modeled gradient-boosting decision tree algorithm. Performance machine learning (ML) assessed AUC values, calibration curves after partitioning dataset 1:1 training/validation. Main results role chance consisting 6 5926), 23920), 1612) included, 3323 embryos. The rate good-quality decreased higher (38.2%, 14.8%, 4.6%, 1.6%, A, B, C, D, respectively; P 0.001). Day earlier lower (Day-5 biopsy: 24.2%, 9.7%, 4.0%, 0.0%, Among biopsied blastocysts, euploidy similar at 44.9%, 43.0%, 42.7%, 66.7% groups A vs respectively (P 0.075, 0.350 0.141). When by number, lowest derived <6-cells, 33.5%, 34.3%, 26.9%, 50.0% 0.803, 0.185 0.394). best performance ML provided % female age, AMH. This resulted modest accuracy predicting development (AUC: 0.72, 95% CI: 0.71 0.73). well- calibrated correctly probability D3-embryos developing blastocyst. For instance, predicted ≤ 2.5% rarely developed (actual rate: 2.6%) had 97.4%.Limitations, reasons caution: retrospective nature inter-observer variability scoring limitation. Nevertheless, embryologists performing followed same SOPs. Findings need validated external cohorts. Wider implications findings utilize estimate an information, mainly when selection PGT-A needs take place. Trial registration not applicable