Since the first “test-tube baby” was born in 1978, in-vitro fertilization (IVF) has been an astonishing game changer when it comes to helping people to conceive. However, as amazing as it is, its success rate still typically hovers around 30 percent. That means that seven out of ten attempts will fail. This can be extremely taxing to would-be parents not only financially, but also mentally and physically. Could A.I. help improve those odds and, in the process, play an important role in the birth of many of tomorrow’s babies?
According to investigators from Brigham and Women’s Hospital and Massachusetts General Hospital, the answer looks to be a resounding “yes.” They are working on a deep-learning A.I. that can help decide on which embryos should be transferred during an IVF round.
“The IVF process involves the insemination of eggs and the culture of embryos externally in a fertility lab before transferring the developed embryo to the mother,” Hadi Shafiee, one of the lead researchers from the Division of Engineering in Medicine at Brigham and Women’s Hospital, told Digital Trends. “A major challenge in the field is deciding on the embryos that need to be transferred during IVF, such that chances of a healthy birth are maximal and any complications for both mother and child are minimal. Currently, the tools available to embryologists when making such are extremely limited and expensive, and, thus, most embryologists are required to make these life-altering decisions using only their observational skills and expertise. In such scenarios, their decision-making process is extremely subjective and tends to be variable.”
An A.I. system used to analyze 742 embryos proved to be 90% accurate at the job of selecting the most high-quality embryo. It does this by evaluating images taken with the microscopes used at fertility centers. While the researchers make clear that this would not be a replacement for human experts, it could help inform decisions that usually have to be assessed manually. (It’s also important to note that IVF doesn’t only fail because of an improperly selected, non-optimal embryo, although this is understood to be a contributing factor.)
As for the next step, Shafiee noted that, “The most important hurdle for such [a] system to be used in clinic is conducting a prospective randomized clinical trial for system validation to pass regulatory requirements.”
A paper describing the work was recently published in the journal eLife.
- Why teaching robots to play hide-and-seek could be the key to next-gen A.I.
- How A.I. bumblebee brains could usher in a new era for navigation
- Your A.I. smart assistant could one day tell if you’re lonely