P-199 Ensuring no clinical risk: a cohort analysis on the agreement between the embryo selected by the embryologist and the embryo selected by artificial intelligence

Abstract

Abstract Study question Is there any clinical risk in selecting the embryo to be transferred using a deep learning-based model? Summary answer Most embryos selected by senior embryologists match the automatically best scoring embryo, and clinical outcomes are better when this coincidence occurs. What is known already In vitro fertilization (IVF) techniques have changed over time with the aim of improving clinical results. Today, embryology is facing a change common to most areas of medicine, the introduction of automation. The use of automated systems in the IVF laboratory is already happening, for example, with electronic witnessing and the ranking of embryos according to their implantation potential. Relying on this type of system is not easy for operators, as they must ensure that the treatments will not be damaged. Study design, size, duration This is a single-center cohort study including 5,411 patients who underwent IVF treatments. Their embryos were cultured in EmbryoScope® time-lapse systems (Vitrolife, Denmark) and routinely evaluated by senior embryologists according to ASEBIR morphological criteria. Then, embryos were automatically scored using the iDAScore v2 algorithm. Participants/materials, setting, methods A total of 7,178 embryo transfers were analyzed. The transferred embryo was selected by the embryologists and retrospectively scored with the deep learning algorithm from 1 to 9.9. Finally, we performed a cohort analysis on the agreement between the embryo selected by the embryologist and the embryo selected by iDAScore v2. The relative risk (or incidence of implantation) (RR), relative risk reduction (RRR) and absolute risk reduction (ARR) were calculated. Main results and the role of chance In general, considering all transfers (fresh and frozen) with known clinical outcome (n = 7,178), the implantation rate was higher when the transferred embryo matched the highest-scoring embryo (58.37% vs. 55.49% p = 0.014). Embryos selected by senior embryologists for single embryo transfer in the fresh cycle (n = 2,783) matched the top-scoring embryo 63.50% of the time. An ongoing pregnancy was achieved in 57.96% of the patients. Out of the 36.54% of patients who had fresh transfer of an embryo that did not have the maximum iDAScore, 553 patients underwent transfer of a single frozen embryo. In this group, the first devitrified embryo coincided with the highest scoring embryo 44.67% of the time (n = 247). An ongoing pregnancy was achieved in 62.34% the patients. However, when the devitrified embryo did not correspond to the highest scoring embryo, the implantation rate was 55.27%. The RR showed that each patient was 1.13 times more likely to become pregnant if the transferred embryo matched the embryo with the highest score. The RRR stated that if there was this coincidence, the probability of implantation was increased by 12.79%. The ARR (0.07) suggested that for every 100 transfers with this match, 7 more embryos would implant than without this match. Limitations, reasons for caution This study is limited by its retrospective nature. Furthermore, the single-center design should be considered when generalizing the results, although our clinic was not involved in the development of this specific model. Wider implications of the findings The high coincidence between the embryologist's decision and the artificial intelligence's decision should comfort assisted reproduction professionals and patients. This coincidence also justifies that realistic models based on artificial intelligence perform the embryo selection procedure as good as the most experienced embryologist. Trial registration number Not applicable