Reproductive biomedicine online · 2025
Never a lost cause: can artificial intelligence help embryologists when only poor-quality embryos are available?
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Gidel-Dissler Nina, Canat Guillaume, Boyer Pierre, Nogueira Daniela, Boussommier-Calleja Alexandra
Аннотация
RESEARCH QUESTION: Can EMBRYOLY, an objectively trained artificial intelligence (AI) system, assist embryologists in embryo assessment when only poor-quality embryos are available for transfer?
DESIGN: Data from 15,767 embryos were collected via EMBRYOLY from 3214 egg retrievals (2019-2024) across 15 clinics (four countries) using three time-lapse systems, including data from seven independent clinics (not used in the original training of the algorithm). EMBRYOLY was used to automatically detect poor-quality embryos. Subsequently, EMBRYOLY's transformer-based model was applied on poor-quality embryos to evaluate agreement with embryologists, ranking performances against clinical pregnancy and live birth outcomes, effect on time to pregnancy and first cycle pregnancy rate. Finally, clinical pregnancy rate was compared between poor versus non-poor embryos recommended for transfer by EMBRYOLY's hybrid model.
RESULTS: For 29% of embryo cohorts, embryologists were faced with only poor-quality embryos available for transfer. EMBRYOLY's first choice of poor-quality embryo was concordant with the embryologists' first choice in 66% of embryo cohorts. EMBRYOLY's score was significantly associated (P < 0.001) with clinical pregnancies and live births on poor-quality embryos. For multiple transfers of poor-quality embryos, the adjunct use of EMBRYOLY could have reduced cycles to pregnancy by 19% and increased first cycle pregnancy rate by 65%. When EMBRYOLY recommended a poor-quality embryo for transfer, it had comparable chances of leading to a clinical pregnancy compared with higher quality embryos.
CONCLUSIONS: Objectively trained AI can help embryologists to select poor-quality embryos that can lead to pregnancy, which is crucial when good or fair embryos are unavailable.
DESIGN: Data from 15,767 embryos were collected via EMBRYOLY from 3214 egg retrievals (2019-2024) across 15 clinics (four countries) using three time-lapse systems, including data from seven independent clinics (not used in the original training of the algorithm). EMBRYOLY was used to automatically detect poor-quality embryos. Subsequently, EMBRYOLY's transformer-based model was applied on poor-quality embryos to evaluate agreement with embryologists, ranking performances against clinical pregnancy and live birth outcomes, effect on time to pregnancy and first cycle pregnancy rate. Finally, clinical pregnancy rate was compared between poor versus non-poor embryos recommended for transfer by EMBRYOLY's hybrid model.
RESULTS: For 29% of embryo cohorts, embryologists were faced with only poor-quality embryos available for transfer. EMBRYOLY's first choice of poor-quality embryo was concordant with the embryologists' first choice in 66% of embryo cohorts. EMBRYOLY's score was significantly associated (P < 0.001) with clinical pregnancies and live births on poor-quality embryos. For multiple transfers of poor-quality embryos, the adjunct use of EMBRYOLY could have reduced cycles to pregnancy by 19% and increased first cycle pregnancy rate by 65%. When EMBRYOLY recommended a poor-quality embryo for transfer, it had comparable chances of leading to a clinical pregnancy compared with higher quality embryos.
CONCLUSIONS: Objectively trained AI can help embryologists to select poor-quality embryos that can lead to pregnancy, which is crucial when good or fair embryos are unavailable.