Summary:

Machine learning algorithms can do a lot of things if they have enough data — recommend products, identify fraud and even help women get pregnant. Fertility startup Ovuline says its 60 million data points from users have helped 50,000 of them get pregnant.

Add another notch in the belt of machine learning. Fertility startup Ovuline announced on Wednesday that its Ovia Fertility mobile app has helped 50,000 users get pregnant in the past 18 months, thanks in large part to all the data its users have supplied for its algorithms to analyze.

Ovuline’s success at predicting fertility is, like so many other things in 2014, the result of smartphones and lots of data, co-founder and CEO Paris Wallace said during a phone call on Wednesday. Whereas traditional fertility studies might have topped out at hundreds of users, he explained, Ovuline’s free app has been downloaded nearly 300,000 times. Its users are adding 1 million data points every two and a half days, or 250 per minute.

“Really, at Ovuline we’re doing the largest study on fertility and pregnancy that’s ever been done,” he said.

Like all things machine learning, the more data Ovuline has about users, the more more accurately it can predict when they’re fertile or, in the case of its new pregnancy app, help guide them through pregnancy. The data also helps it provide instant feedback, which is something many apps and other quantified-self efforts fail to do. If, for example, a woman notes that she’s having a particular ache, the Ovuline app can tell her whether it’s normal and what percentage of other women also have it during this phase of their pregnancy.

From a data perspective, I think the really interesting part is how Ovuline goes about acquiring the data it needs to perform its analyses. It can gather data from other apps such as Jawbone and Fitbit, it gamifies entering certain medical records and historical data, and women are willing to share other personal information about their bodies, emotions and even sexual activity.

“There’s no quantified-self app that’s telling us things like intercourse or cervical fluid quite yet,” Wallace joked. “We’re still waiting on that.”

slide-3

Ovuline’s approach, and success, underscores one of the big trends of our upcoming Structure Data conference, which is that if you need data about something in this day and age, there’s probably a way to get it. Sometimes, that technique might be good, old-fashioned direct questions. If users perceive a real benefit will accrue from it, they’ll probably answer.

And in the case of fertility, there is a real benefit. Wallace said a recent, respected study showed that 40 percent of women who undergo fertility treatment are not really infertile, and he noted that many fertility treatments are really just methods of amplifying intercourse (e.g., helping a woman release more eggs, or fertilizing an egg outside the body first).

For most women, he said, “If they knew when they were ovulating, they could get pregnant.” Heck, he added, 10 percent of Ovuline’s medically infertile users have been able to conceive while using the app.

At this point, Wallace is also playing the diplomat when it comes to competition from other fertility apps such as Kindara, Clue and the Max Levchin-backed Glow. With millions of pregnancies and attempted pregnancies each year, he thinks there’s plenty of opportunity to go around.

The problem isn’t people going to the app store and deciding between which app is best, he said, “The problem is people coming to the app store trying to get pregnant.” That the options exist and seem to work is testament enough to how powerful data can be.

Feature image courtesy of Shutterstock user Guskova Natalia.

Comments have been disabled for this post