In vitro fertilization (IVF), as it is now carried out, depends on skilled embryologists visually evaluating and rating the embryos to decide which should be implanted. Since the inception of IVF, this procedure has been the gold standard; nonetheless, it is widely recognized that it has limits. The difficulty of embryologists to reliably reproduce embryo selection is the main cause for concern. Although the exact effect on success rates is difficult to predict, it makes sense to believe that a more standardized procedure could raise success rates. Some academics are investigating whether artificial intelligence (AI) may be used to more consistently choose the “healthiest” embryos for implantation in light of the current expansion of AI and its applications in the healthcare industry.
Artificial intelligence: what is it?
A group of technologies known as artificial intelligence use vast amounts of data to teach a computer to do activities that would normally need human intelligence. One of these technologies that is frequently used in the healthcare industry is machine learning. It creates models based on patterns discovered in vast volumes of data using statistics. Artificial neural networks, which are designed to resemble how the human brain processes information for use in decision-making, are used in deep learning, a more sophisticated kind of machine learning that has potential applications in in vitro fertilization (IVF).
The purpose of utilizing AI in IVF embryo selection is to increase the confidence of the embryologists’ decision-making process through a data-driven method, hence removing the potential bias associated with depending exclusively on visual assessment. The machine learning method can help choose which embryo will be most robust for implantation by comparing the embryo against a set of desirable qualities using time-lapse photography, which is used to take photographs at regular intervals over the 5-day observation period.
Dr. Daniella Gilboa, an Israeli embryologist and co-founder of AIVF, is one of the pioneers in this field of study. She discussed her attempts to advance this technology in IVF in a recent Moneyball Medicine podcast interview. Dr. Gilboa and her associates are working with research institutions around the world, including as Stanford University, Cornell University, IVI in Spain, London Womens Clinic in London, and Embryo Lab in Greece, to create an AI-powered end-to-end decision assistance tool for embryologists.
Their patented software examines every embryo image and evaluates millions of characteristics that are combined to determine success (defined as implantation). The likelihood of each embryo’s success would then be visible to the embryologist. By providing the patient with access to the same platform through a patient portal, Dr. Gilboa intends to engage the patient in the process. This will enable the patient to view their embryos, as well as the probabilities, and discuss embryo selection with the embryologist from a more knowledgeable position.
Industry is not the only group conducting this research. An organization at Cornell University recently shared their experience of teaching a Google deep learning tool that was available “off-the-shelf” to categorize embryos into “good, fair, or poor” categories according to how likely they were to implant. There was minimal consensus among the embryologists when they were asked to assess a portion of the embryos used in this investigation. Nonetheless, the researchers obtained a majority vote for each embryo from the group of embryologists through the use of a voting system. Their AI algorithm predicted the majority vote with a precision of 95.7%, suggesting that the system would be more adept at choosing the best embryo than any one embryologist.
Dr. Gilboa believes that this technique is needed for reasons more than only selecting embryos more carefully in order to increase the success rate of IVF. “IVF is evolving into something new these days,” she explains. IVF is not something you undertake just when advised by a doctor.If it’s optimized and the current course of treatment wasn’t as harsh or severe…The need for in vitro fertilization will continue to rise. Utilizing AI technologies is the only method to do the task correctly.” This is the main motivation behind her job, in her opinion, as she helps more families become pregnant in less time and with fewer complications.
Before AI technology is widely used for embryo selection, further research is still required, but this is a potential advancement in the field that could increase both the technology’s accessibility and success rates.