Retrospective Evaluation Blastocyst Development by Artificial Intelligency

Overview

In assisted reproductive technology (ART), selecting the most viable embryos from a large number of fertilized eggs is crucial. While techniques like morphological assessment, time-lapse monitoring systems, and pre-implantation genetic testing have improved the process, implantation success rates remain limited. The introduction of artificial intelligence in embryo evaluation, such as automated systems like EMA by AIVFTM, provides a promising alternative to enhance the accuracy and effectiveness of embryo selection. This study aims to assess the performance of the EMA system compared to traditional methods by examining its ability to rank embryos based on their potential for successful implantation, with the goal of increasing the chances of a successful pregnancy. This is the first clinical evaluation of this platform in France, offering new opportunities to improve decision-making in in-vitro fertilization.

This monocentric study analyzes 420 ART cycles from the Clermont-Ferrand University Hospital, conducted between January 2022 and December 2023. It includes IVF/ICSI cycles with embryos cultured in a Time-Lapse Incubator (Geri) and both fresh and frozen blastocyst transfers. A total of 1211 embryos were analyzed, with 619 used to train algorithms for predicting reproductive outcomes. This included 551 single and 34 double embryo transfer cycles, divided into 274 fresh and 345 frozen blastocyst transfers. For all blastocyst-stage embryos, the Geri score, Gardner classification, and EMA score based on time-lapse video were recorded for statistical analysis and evaluation of reproductive outcome predictions

Conditions