Testing an artificial intelligence algorithm to predict fetal heartbeat of vitrified-warmed blastocysts from a single...

Area of Science:

• Reproductive Medicine
• Embryology
• Artificial Intelligence

Background:

• AI has proven effective in assessing fresh embryo quality.
• Predicting outcomes of frozen embryo transfers remains challenging due to potential post-warming complications like collapse or degeneration.

Purpose of the Study:

• To investigate if an AI algorithm can predict fetal heartbeat from images of vitrified-warmed embryos.
• To assess the AI algorithm's potential in improving frozen embryo transfer success rates.

Main Methods:

• A retrospective cohort study analyzed 1109 embryos (568 vitrified-warmed, 541 fresh controls).
• AI algorithm analyzed four types of time-lapse images (pre-vitrification and post-warming).
• AI scores were correlated with conventional morphology and fetal heartbeat outcomes, analyzed via logistic regression and ROC curves.

Main Results:

• AI scores from pre- and post-warming images significantly correlated with fetal heartbeat.
• The AI score showed predictive significance in oocyte donation and autologous oocyte cycles.
• AI scores varied by morphological category, differentiating viable from non-viable embryos post-warming.

Conclusions:

• AI algorithms show promise in predicting fetal heartbeat from vitrified-warmed embryo images.
• This AI tool could assist embryologists in managing frozen embryo transfers, potentially reducing cycle cancellation rates.
• The AI algorithm, requiring only a single photo, is adaptable for clinics with varying incubator systems.