2022 Morbeck

Artificial Intelligence in Embryology and ART

While developments in AI, machine learning and automation were covered extensively at this meeting, a direct debate concluded that there is still a long way to go before machine learning algorithms live up to expectations, and that more scientific, ethical and practical research is needed on AI.

This conference debate on artificial intelligence in the lab reached a consensus – that while there are benefits in the application of AI in ART, there is still insufficient evidence to support its widespread use.

The first to speak was Dean Morbeck, Chief Scientific Officer of Kindbody, US, who argued for machine learning and AI-guided decision making on the basis that it improves embryology and fertility treatment. He outlined key benefits such as improved and standardized embryo classification, more efficient and accurate data entry, a better ‘patient experience’, greater lab efficiency and automated quality control.

The human brain, he said, couldn’t possibly handle everything involved in developing embryos. Even when assessing the likelihood of implantation, embryologists are dealing with myriad variables and “the brain has no way of doing this.”

Properly arranging embryos is the most important decision an embryologist will make, Morbeck said, but humans don’t have the ability to be objective when judging and have “inherent subjectivity”; this is also an issue that affects decisions about whether or not to freeze blastocysts or biopsy them. The staff may be overworked, making decisions only because it’s a weekend, or just tired.

This, then, is one area where AI could help increase the chance of live births, Morbeck argued. Clinics fail more than they succeed in IVF, clearly demonstrated by SART data in which 14% of cycles produce no embryos and 62% no live births.

Patients who have not been successfully treated leave with ‘nothing to prove’. To improve the situation, Morbeck explained that machine learning tools could standardize the “minimum usable threshold” applied to reduce the rejection of viable but low-grade embryos. AI could learn how embryos develop and then use that information to select the best embryo.

The case against AI was brought by Peter Tennant, associate professor of health data science at the University of Leeds, UK. His argument was that much of the hype around what AI can do is based on ‘overpromise’ and that the technology still has a long way to go to prove its worth in embryology and ART. To illustrate, he said Google’s medical AI — such as its algorithm to detect diabetic retinopathy — has proven accurate in the lab, but not in real life.

Specifically in embryology, Tennant outlined the promise and problems of AI. Some of the benefits include improvement and automation (e.g. follicles counting), prediction and classification (e.g. live births), smarter and more personalized care, and targeted treatment regimens. However, the problems are significant: machine learning is developed in small and specialized samples and contexts, performance is variable, and external validation is lacking. AI-based tools also rarely focus on a really interesting outcome, Tennant said. The few studies that have focused on live birth outcomes have shown that AI performs poorly.

While it may seem like a good prediction and prediction tool, AI has significant limits, including a lack of common understanding: Algorithms only see patterns in data and so can be fooled by a doctor’s purple pen mark on a cancer biopsy slide. Tennant said this could happen in AI analysis of blastocyst images — an algorithm can “see patterns that we don’t want to pick up.”

In response to a question from the public, Tennant acknowledged that humans make mistakes too, but emphasized that a skeptical approach is needed around AI until the technology is fully understood, ending the session by saying it’s not a matter of fear. for intelligent machines , but from stupid ones.

Leave a Comment

Your email address will not be published.