Putting eggs on ice and In vitro fertilization (IVF) options are growing in popularity with millennials, but these techniques don’t guarantee a successful pregnancy. Data from one of Europe’s largest IVF clinics shows most eggs survive the freezing process, but only about a third result in successful pregnancies.
However, a Canadian startup is hoping to change that.
Currently, fertility doctors use a rather old method for determining the odds of a woman’s pregnancy: a calculation based on age, eggs retrieved and historical averages. Toronto-based Future Fertility thinks there’s a better way, using artificial intelligence.
Future Fertility’s co-founders claim its neural network, Violet — named after one co-founder’s daughter — can successfully predict fertilization 90 percent of the time based on a single image of an egg.
While that’s a good start, Violet isn’t as good when it comes to predicting other important IVF milestones. For example, Violet is only accurate 65 percent of the time when it comes to assessing whether an embryo will survive at least five days. The AI’s ability to predict successful implantation in the uterine wall is even less accurate.
However, Future Fertility says those numbers will get better when Violet rolls out to IVF clinics, where it can learn from every patient going through the process.
Future Fertility isn’t the only one experimenting with AI and IVF. Researchers from Cornell University-trained an off-the-shelf Google deep learning algorithm to grade the quality of embryos.
Future Fertility trained Violet using 20,000 images from TRIO
Violet was trained using 20,000 images and de-identified electronic health records from Toronto’s TRIO fertility clinic, along with a handful of partner IVF clinics. The neural network used the files to learn the characteristics of eggs that fertilized versus ones that didn’t — an impressive feat, considering that to the human eye, eggs look like formless spheres of water.
Once trained, Future Fertility validated Violet using roughly 2,000 images withheld from the original training set.
However, the company hasn’t published any of its results in a peer-reviewed journal yet.
Despite that, Future Fertility is planning a trail that will likely enroll a few hundred IVF patients. Violet will score eggs from half the participants based on photos taken during the few hours after they’re harvested. The AI will also predict each egg’s chances of IVF success.
The other half of participants will go through the standard calculation process, and an embryologist will try to make predictions about each egg’s fate.
After, Future Fertility will follow the patients and compare fertilization, implantation and birth rates to see if the AI is more successful than traditional methods.
The company could launch commercially later this year
Additionally, the company has agreements to begin testing Violet with seven IVF clinics, along with TRIO, in the U.S., Japan and Spain.
Future Fertility suggests this is a beta phase, and hopes for a full commercial launch later this year.
The company ships a system including a small camera that attaches to a standard light microscope for taking pictures of eggs, and a software package to upload images to Future Fertility’s servers so Violet can scan and score them.
As for monetization, it’ll be up to individual clinics to decide whether to seek a subscription to incorporate egg scoring as a standard part of egg freezing and IVF or offer it as an add-on service with additional fees. Even if it cost more up front, it could save patients money in the long run by letting them know earlier if they’ll need to look into alternatives.
This article originally appeared in Mobile Syrup