The baby that Ashley Sullivan is carrying is one she watched closely from the very start, and it's one that a computer and an embryologist also watched closely. Sullivan, who is due in September, had a long path to pregnancy. She and her husband, Patrick, tried getting pregnant for more than two years before they saw fertility doctors and learned they would need in vitro fertilization.

During their first round in December 2017, they retrieved 28 eggs, but only five formed blastocysts when fertilized, and only four were chromosomally normal when tested. She had to rely on information through phone calls and hard-to-understand voicemails about what was happening to her embryos.

It was an emotional roller coaster that continued once an embryo was chosen and implanted. It didn't work.

They waited a month, picked another embryo and did a second implantation. At day nine after the implantation, hormone levels weren't good, but something was happening. By day 12, she was told it was over.

"After that, I was at a loss," she says. "What's wrong with my body? Why is it fighting off and not accepting these embryos?"

Another egg retrieval in the fall resulted in 22 eggs. This time she was able to see the video of her embryos turning into blastocysts as part of a study being done by Ovation Fertility. 

Daily emails included updates and videos. She didn't have to try to remember when a phone call was scheduled or worry about missing a phone call and trying to decipher a voicemail.

"Having the emails and the videos made a huge difference," she says. "You feel like you were a part of it."

And she learned that on day six, she had one blastocyst, and it was normal. That was the magic embryo.

What Sullivan didn't know was that, along with the video study, her embryos also were part of an artificial intelligence study being conducted by Ovation Fertility. In addition to an embryologist looking at the potential blastocysts, a computer at the Australian company Life Whisperer was looking at and grading them.

The problem with having a human pick embryos to implant is "it's very subjective," says Matthew "Tex" VerMilyea, the IVF lab director at Ovation Fertility. Each embryologist could grade an embryo with a different letter on the A-F grading scale, which embryologists have tried — but not successfully — to make universal. The scale looks at the tightness of the cells and how many cells are forming, among other things. "There's no standardization," VerMilyea says.

All the ones that the embryologist might select as looking healthy might not translate to being chromosomally normal. Yet, an embryologist will rank them and choose the one he considers AA first, then BB, then CC. They try not to go below CC.

With Life Whisperer, VerMilyea can upload photos of the blastocysts and, within a few seconds, the computer grades them. He can sort them from the most viable to the least based on percentages. Life Whisperer is comparing the blastocysts to its more than 10,000 file photos, with the knowledge of which of those file photos became a successful pregnancy and baby. 

In a blind study, the computer had a 32 percent increase in selecting the best blastocyst compared with the embryologist — from 60 percent to 92 percent.

The hope is that using a computer will mean that doctors no longer have to order genetic testing for each blastocyst. That testing typically involves sticking a needle into the blastocyst and removing a cell to analyze. 

"This is noninvasive," VerMilyea says of using the computer.

VerMilyea says that while more testing on artificial intelligence needs to be done, he sees it as something embryologists will use when they are considering between several embryos that they might have identified as being the best.

"Hopefully, it will become commonplace," he says.

Dr. Kaylen Silverberg of Texas Fertility Center, which uses Ovation Fertility as its lab, says his patients have loved being part of research like this study and the video study. As well as improving accuracy, he says he believes using a computer to pick the embryo will save money and protect the integrity of the embryo.

Silverberg says for people who have had embryos with chromosomal abnormalities, using genetic testing improves their rate of pregnancy because it rules out embryos that are not normal; for those who have normal embryos, doing genetic testing can lower their pregnancy rate because of the risks to the embryo from the needle and from freezing it while waiting for the results.

Right now the second phase of the study is looking at whether using AI is superior to chromosomal testing.

"We've got to make sure it works," Silverberg says.

He expects they will be met with skepticism by fellow embryologists and fertility doctors, as well as concern that fertility companies might lose revenue. He expects this could save $5,000 in every case that uses AI for picking the right embryo and for genetic testing.

"The goal is the highest pregnancy rates and to lower the cost of a single healthy baby," Silverberg says.

It will take more research and a change in thinking. "We are light-years from making it standard," he says, but he is hopeful.

Sullivan says she believes that if AI had been used to pick an embryo in the first two rounds, it might have selected a different one. She now knows those embryos were considered CC by the embryologist.

"I really think this could be a game changer," she says.