An artificial intelligence (AI) system selected the same embryos as an embryologist 69 percent of the time, with no significant difference in clinical pregnancy rate.
Researchers from 14 clinics in Australia, Denmark, Sweden and the UK compared pregnancy rates and outcomes from embryos selected using a deep learning AI system, iDA, to those selected using standard assessment of the morphology of the embryo by an embryologist. Though no significant differences in pregnancy rate were found, the selection of the embryos using AI was ten-times quicker than selection by embryologists.
'Deep learning may have many medical applications, but this is so far one of only a few prospective randomised trials of the technology in any area of healthcare,' Dr Christos Venetis, a senior research fellow at the University of New South Wales in Sydney, Australia and one of the study authors, wrote in The Conversation.
He added: 'It seems from this that the use of a deep learning tool for embryo selection will not radically change the outcome (as it mostly chooses the same embryo) for a patient undergoing IVF,' said Dr Venetis. 'However, the use of a reliable automated tool of this sort may make embryology laboratories more efficient and consistent.'
During IVF treatment an embryologist will review a patient's cohort of usable embryos to determine which to transfer into the patient's uterus, based on which is most likely to lead to pregnancy. This selection is made on the basis of their structural features, termed morphology. Additional information such as the timing to reach key developmental events can be gained from using timelapse incubators, used by some clinics, where video footage is acquired while the embryos are developing.
The process of reviewing the resultant 'morphokinetic data', however, is time-consuming and can be subjective, with embryologists not consistently selecting the same embryo for transfer. A number of different AI systems have been developed to identify patterns between successful features in embryo morphokinetic data and aim to be able to improve the process of embryo selection.
In this latest paper published in Nature Medicine, iDA was used to review the same morphokinetic data as an embryologist, and to score each embryo out of ten. The embryo transfer procedures were then randomly allocated to the two arms of the study of 1066 patients: with half having the embryologist-selected embryo transferred, and half the embryo which scored highest using the AI system.
The clinical pregnancy rate (the detection of a fetus on ultrasound) was compared between the two groups. The results were not significantly different, with the AI system slightly lower at 46.5 percent compared to 48.2 percent in the embryologist-selected group. In fact, the AI system and the embryologist selected the same embryo more often than not, in 68.5 percent of cases.
Due to the challenges of conducting larger randomised control trials, Dr Venetis said researchers were looking at different approaches to determine the applicability of the AI system they had tested.
Sources and References
-
Deep learning versus manual morphology-based embryo selection in IVF: a randomized, double-blind noninferiority trial
-
Can AI pick IVF embryos as well as a human? First randomised controlled trial shows promise
-
Artificial Intelligence can pick best embryos for IVF faster than scientists can, study finds
Leave a Reply
You must be logged in to post a comment.