Skip to main content

This startup will tell you how much experience a doctor has treating your condition

This startup will tell you how much experience a doctor has treating your condition

/

Amino analyzed health insurance claims from 188 million Americans

Share this story

If you buy something from a Verge link, Vox Media may earn a commission. See our ethics statement.

Choosing a doctor can be a real pain. Besides looking at a physician’s credentials — where they went to school, for instance — or checking for online reviews that may not be truthful, there aren’t many resources that can help someone decide if doctor A is more experienced or runs a better practice than doctor B. That’s the problem that Amino, a new website launching today, wants to solve. By analyzing data from 3.9 billion unique health insurance claims, the company thinks it can help people figure out if a doctor is experienced enough to treat a given condition, or even if they’re any good.

No reviews, just insurance data

It’s an interesting idea, but the data that Amino works with — health insurance claims — may make some people uncomfortable. And because the startup’s search results come in the form of numbers and bar graphs, instead of patient reviews, Amino might have to work hard to convince its audience that the information it’s presenting them with is actually useful. Because Amino won’t just have to compete with companies like Zocdoc and Healthgrades; it’s also going to have to teach consumers a whole new way of evaluating their physicians, too.

When a user searches Amino’s database, they’re asked all the questions one might expect to answer when searching for a doctor online. Users must name the kind of doctor they want to see, their health insurance company, their age, and their gender. But instead of showing patient reviews, Amino draws on health insurance claims to give users an idea of how experienced doctors in a geographic area might be. A search for "celiac disease" in Manhattan will return the names of doctors who treat patients with that condition most often — a proxy for experience that Amino says correlates with favorable outcomes. And a search for "mammogram" will do the same, but with the added bonus of letting users know how long it might take to get a biopsy following abnormal results.

"We don't have permission to name the sources at this time."

"It's their experiences with nearly 900,000 physicians in the US," says David Vivero, CEO of Amino and founder of Rentjuice, a company later acquired by Zillow. The company points out that it has health insurance claims from 188 million Americans on file. "The data is anonymous, but it allows us to see where those folks have gotten care."

Thanks to partnerships with insurance companies, employers, health IT companies, and large health care providers, Amino has gathered data from 3.9 billion unique health insurance claims. The company isn’t sharing who their partners are at the moment, but that could change, says Iz Conroy, communications manager for Amino. ("All of the data is permitted to be used in Amino's products and services, but unfortunately we don't have permission to name the sources at this time," she told The Verge in an email.) So, when people search Amino’s website, they’re searching through billions of health records — records that may or may not include their own. That’s the information that Amino wants its users to think about when they evaluate a physician and book appointments through the site, free of charge.

The launch of a company like Amino "isn’t surprising," says Jay Wolfson, a public health researcher at the University of South Florida. It’s the next logical step from what products such as WebMD and HealthGrades offer — and that’s something people have been talking about for some time, he says. "I think this is a direction that we have all been wanting to go in." But it’s not enough for Amino to say that they’ve gathered and modeled all this data, because the databases from which Amino gleans its information "are unclear," Wolfson says — and that could be a problem. Right now, Amino has a lot to prove.

Testing Amino

I had the opportunity to try a demo version of the service, and it worked reasonably well. When I searched for "migraines," the website showed me where a woman around my age in Manhattan might be able to find an experienced neurologist. But other queries didn’t work as expected.

Room for improvement

A search for a female "general practitioner" who treats 30-year-old women with my health insurance, for example, turned up two doctors in my area — neither of whom were my actual doctor. But when I used the term "family practitioner" with the same criteria, 25 different doctors popped up, including my own. People who know that a general practitioner is distinct from a family practitioner will have no problem navigating Amino’s site, but for everyone else, that could be an obstacle. Competing website Zocdoc, on the other hand, has already figured out that lumping certain doctors into the "primary care physician" category can prevent this distinction from muddling search results.

Given my experience with the site, I’d say that it’s easiest to use if a patient already has a diagnosis, or if she knows exactly what type of specialist she needs to see. For those people, Amino really could bring some clarity to their health care decisions. But attracting that audience might prove challenging, if only because a lot of people who find themselves in that situation opt to see the specialists that their primary care physicians recommend.

Amino could add clarity, but attracting an audience might be hard

It’s also unclear how users will respond to a website that focuses so heavily on numbers, while providing little interpretation. Should a user favor a doctor who has treated 15 people for a given condition in the past four years when other doctors in the area have seen five? And how important is it that only 1 percent of patients that a doctor sees have a sprained ankle? A data-driven approach might be refreshing to some, but the lack of patient reviews — which, unlike health insurance claims, can be an indicator of bedside manner — might turn others away.

Vivero doesn’t seem too worried about how people will react to the graphs presented on Amino. "People want an unbiased set of information to help them feel good about their decision," he says. "If we can give people more and more facts, we think that they can at least make a decision confidently on their own — as opposed to making a decision which maybe wouldn’t align with their values if they had all the data in front of them."

"People want an unbiased set of information."

Wolfson agrees; Amino could appeal to a lot of people, he says. "The baby boom generation after World War II has been very consumer-oriented and the generation following us is far more tech-savvy, and we are used to consulting computers to query everything from the price of gasoline to the options for hotels and airline tickets," he says. "So I think some people would be interested in this."

Unfortunately, until Amino releases more information about the data it’s using, evaluating the quality of its service will be hard. Data scientists introduce judgments into their analysis all the time, simply by opting to use one data source over another, says Jennifer Miller, a medical ethicist at New York University. So by making deals with one health insurance company, but not another, Amino might be missing some important piece of information that can’t be identified right now. "So, what we need to do is see if Amino’s corporate judgments align with the way in which patients understand quality services," she says. "People might understand what a quality apartment looks like in a service like Zillow, but defining quality in health care is much more complex."

Increasing transparency might help

When asked about this, Vivero answers that the data that powers Amino "isn’t static." The company plans to add other data sources in the near future — sources that will help Amino get the "clearest picture possible" of a doctor’s practice. That’s probably a good thing, but if people don’t know what those additional sources are, that won’t mean much.

Then there’s the question of security. The de-identification process happens before the data hits Amino’s servers, and the startup’s partners prohibit the company from re-identifying the data, Vivero says — but analyzing private health information belonging to 188 million Americans is still a big responsibility. That means that Amino will have to find ways to reassure the public that the company’s data won’t be re-identified somehow, and that the startup’s security protocols are top-notch. "I think the incentives for re-identifying people in a database can be low," Miller says, "but you know that could change in the future."

Amino raised $19.4 million in its latest round of funding

Right now, users can search Amino’s website free of charge, and the site doesn’t feature any ads. Vivero explains that this is meant to show that people can’t use money to influence or limit the results that appear on Amino. The company raised $19.4 million in its latest round of funding, which will help keep Amino running this way for a while, Vivero says. "Our belief is that we'll figure out a business model around our consumer products over time." And in the meantime, Amino plans to make money by selling part of its analysis to private companies, he says.

Today’s launch is just the beginning for Amino, Vivero says. "We'll be doing lots and lots and lots of different products built off of the data set, because as you can see, you could build all sorts of products that focus on different decisions that people make." The company also plans to increase the amount of information included in its search results, which could add value to an already interesting product. Of course, that increases the pressure on Amino to show that all that data really can lead to better care.