This start-up has stumbled across an incredibly popular use for the Apple Watch

Key Points
  • The app, called Cardiogram, said it has 100,000 daily active users and a 73% open rate
  • The app can also be used to screen for heart rhythm abnormalities
  • This is the latest in a series of studies that are exploring how artificial intelligence can root out abnormalities in medical data
A signage for the Apple Watch is displayed in the Apple store in Saint Germain, Paris, France on December 1, 2016.
Chesnot | Getty Images

The Apple Watch is proving itself to be a useful tool for monitoring serious medical conditions, and not just for fitness.

A mobile health app made by start-up Cardiogram, backed by Silicon Valley venture firms including Andreessen Horowitz, has proven to be shockingly popular among the people who download it, the company's co-founder Brandon Ballinger told CNBC. Although only 250,000 people have downloaded it to try it, more than 100,000 people use it every day. Of the people who kept the app, more than 73% open it every single day.

Cardiogram uses the Apple Watch's built-in heart sensor to give you advice about heart health -- for instance, showing when you're unusually stressed out by checking your heart rate.

The company is also doing original research, and on Thursday it unveiled the results of a study on Thursday showing how the Apple Watch can be used to detect abnormal heart rhythms.

The company is presenting its results at a conference for cardiac electrophysiologists, called Heart Rhythm Society.

"Cardiologists don't have time to look at streams of data flowing in from [Apple's] HealthKit and Google Fit," said Ballinger, who formerly worked at Google before starting Cardiogram. Both Apple and Google have developed software systems that are designed for users to safely store and selectively share their health information, mostly collected via sensors. "That's not a task that humans do, but machines can review all this data."

The company developed a neural network — a computing system that learns from analyzed data — which it trained on millions of its users' heart rate measurements. The goal was to identify atrial fibrillation, which is an irregular heart beat. The study found that it's accurate 97% of the time using the smartwatch's heart rate sensor, compared to screening tests performed at the hospital.

For the study, it partnered up with UC San Francisco's Health eHeart study, which is using mobile tools to collect a massive data-set of heart health data from 1 million people.

"Atrial fibrillation is one of the most common heart rhythm disturbances and a common cause of stroke," said Greg Marcus, a cardiac electrophysiologist at UC San Francisco. "But a lot of people don't realize they have it."

"The hope in working with Cardiogram is to leverage tech that people are wearing anyway to identify those with atrial fibrillation or at high risk for it," he added. In these cases, users might use a mobile medical device like an AliveCor to screen for potential problems, or contact their doctor.

This is the latest in a series of research studies finding that artificial intelligence can be a useful tool to assist physicians in screening for medical abnormalities. Google's Brain team has also conducted research into a variety of ways to use artificial intelligence to detect subtle indicators in medical scans.