Is the gory 'Instagram for doctors' the future of medicine or an overhyped start-up?

By Fusion Staff | October 19, 2015 | 11:00am

Latest

Figure1 is like a little shop of horrors. The Instagram for Doctors has gnarly photos of scars, brain scans, bloody organs, and body parts that are oozing and green.

When Joshua Landy was studying medicine at Stanford, he saw doctors taking pictures of patients’ wounds or infections and sharing them with each other via text or email. It wasn’t usually just for fun. Often, they were asking each other for help on cases that stumped them. But this social sharing through unencrypted channels was problematic due to the strict privacy laws that protect people’s health information so Landy decided to create a service, Figure 1, that would facilitate sharing while keeping patient information private.

Most Popular

The company strips uploaded photos of geolocation data and uses the same security standards to keep them safe that hospitals must use to safeguard medical records. It has easy tools for blacking out identifying information, like people’s faces or tattoos. Launched in 2013, it now has more than 500,000 members—verified doctors, nurses, and healthcare workers who share and comment on the strange, at times hard-to-diagnose cases they encounter on a daily basis. The rest of us can look, but not partake in the discussion.

There’s been stories of the app helping docs crack difficult cases, such as when a doctor posted a photo of the hands of a patient whose nails were missing and deformed. A fellow user suggested it might be a rare genetic condition dubbed Naegeli-Franceschetti-Jadassohn (NFJ) that affects one in every 2 to 3 million people; amazingly, the user was right. In another case, a physician assistant in rural Oklahoma was able to detect a severe case of shingles meningitis, because she’d seen images of serious shingles on the app.

Based on these anecdotes you might think Figure1 is the future of medicine. It seems incredibly promising after all, putting half-a-million medical minds to work on health issues that stump us. Despite several articles hyping the site—some outlets have said it’ll help doctors “‘figure’ out your condition in a jiffy“—there’s no solid evidence yet that the app is improving medical diagnoses or the way that care is delivered, or that it ever can. Medical experts told me Figure1, and apps like it, might be good for solving “zebras,” the term doctors use to refer to super-rare cases, but they were also quick to say crowdsourcing through the internet won’t revolutionize healthcare.

“You know we make a big deal, in part because we all watch a lot of television and you see the Dr. House diagnostic dilemmas, but in my experience, they don’t happen that often,” said Dr. Robert Wachter, a professor of medicine at the University of California, San Francisco and the author of The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine’s Computer Age. “I think the more important use of IT (information technology) will be an easy path for me as a primary care doctor taking care of a patient to get the opinion of a [specialist].”

Figure1’s Landy told me that only five percent of the app’s users are uploading photos. Thirty percent actively comment leaving messages like, “So pretty but so dangerous. This is #Pseudomonas.” The rest are simply lurking, maybe learning or maybe just voyeuristically consuming images of patients’ weird ailments the way we consume our friends’ vacation snapshots on Instagram.

Even though Figure1 has tools that obscure patients’ identities, some hospitals have clear policies against staff taking pictures of patients with cellphones. Others don’t, but physicians are still leery of contributing to it.

“I don’t upload to it largely because of my own concerns about being at a large institution where rules around that are extremely vague,” said Dr. Satish Misra, a cardiology fellow at Johns Hopkins University.

Figure1 isn’t the only medical startup to go the crowdsourcing route. There’s CrowdMed, a service that lets people upload photos, videos and descriptions of their own unsolved medical cases so that a crew of “medical detectives” can create a list of possible explanations and treatments, an unofficial diagnosis the patient can then take to their physician. (Currently, Figure1 doesn’t allow patients to upload images of their own scabs and scars.) Doximity, the LinkedIn for Doctors, also lets physicians collaborate on cases online. It boasts more than 500,000 members. On HealthTap, doctors can rate drugs so patients can zero in on medications that might work for them.

All these apps face similar obstacles: the potential legal toxicity of the information they’re making more sharable. In other industries—recommendation sites, social networking, even transportation—data is easy to come by. So companies like Yelp, Google, Facebook and Uber have been able to build empires on the shoulders of the masses. But health care is a thorny beast. Data is heavily protected, hard to come by and requires delicate treatment once amassed. These are considerations crowdsourcing services like Figure1, which are working from within the health care system, must take into account. That puts them at a disadvantage before they even get to the starting blocks.

Figure1 can only provide its community of medical sleuths a myopic view of a patient’s condition. None of their medical history or identifying information is attached to the photo of their current ailment. That means a physician weighing in on a case lacks important information necessary to make a valid diagnosis, a point Landy concedes, and why Figure1 comments are meant more as suggestions rather than air-tight solutions. If marketed as such, the company would also have to contend with prickly state licensing laws for doctors that limit where physicians can dole out professional advice. IBM, by contrast, can feed its Watson supercomputer millions of images and full medical record to help doctors come up with an actionable diagnosis locally.

Beyond privacy, another problem with medical crowdsourcing is that it takes time. And when you’re trying to crack a medical case, time is your worst enemy. Even if someone across the world has the nugget of knowledge you need, it doesn’t mean they’re going to get it to you in time. “That’s where crowdsourcing really fails,” says Dr. Iltifat Husain, the founder of iMedicalApps.com and an assistant professor of emergency medicine at Wake Forest Baptist Health in North Carolina.

Figure1 is trying to address that by finding the person in the crowd who’s best suited to make a diagnosis, and trying to get them to comment as quickly as possible—that “easy path”—to a specialist that UCSF’s Wachter alluded to. In April, it introduced a feature called Paging that pings the medical professional on its platform who its software has determined is best positioned to diagnose a difficult case, based on factors like their specialty and time zone.

Say, for instance, that a doctor needs help diagnosing a weird lump in an abdominal x-ray, but there’s no local specialist with the right know-how. He can upload the image to the app, along with a brief description. Figure1 will then page a specialist in their database to take a look and provide feedback. If that person is unavailable, the system pings another, until somebody responds. It takes about 30 minutes to get a response, which is less than the time it can take to get a response even if there was someone else working within the same hospital, according to Landy. In the future, Landy wants to build more powerful machine intelligence to help computers triage Paging queries to specialists who give the best and most useful responses and who are correct most often. Theoretically, there could be value in that.

The problem, though, is that Figure1 is totally unconnected,” says Dr. Bob Kocher, a partner at venture capital firm Venrock and a former special assistant to the President for healthcare on the National Economic Council. What he means is that Figure1 lacks any of the detailed patient history that would make Figure1 photos a really valuable medical resource. And that’s, again by design, because otherwise Figure1 could never legally get away with having doctors post images of patients.

For now, Paging is free, but Landy thinks it’s one of the ways Figure 1, which currently isn’t profitable, might be able to make money in the future. Digital healthcare companies like GrandRounds are building lucrative businesses around second opinions provided over the internet. That startup has raised a total of $106 million from investors. But, GrandRounds has a huge leg up in that it’s integrated into the health system. GrandRounds doctors get a patient’s medical records, which they use to make an official diagnosis. That service is covered by insurance. An answer to a Figure1 Page is a mere suggestion. It’s not actionable, which in the health care world means insurance companies won’t pay for it. That, Kocher says, is probably Figure1’s biggest obstacle. Figure1 will need to work around this, or figure out another revenue strategy—perhaps advertising. It’s easier to serve an ad, than serve a patient.

And so, behind the scenes, it’s working to build a database, according to Landy, that will let help Figure1 make better sense of the streams of disturbing images coming in. It’ll use that to surface the right content to people; a neurologist, for example, might not want, or need to see, dermatology photos. Landy thinks medical vendor advertisers would be interested. Every day, Figure1 users see 6 million images, Landy said. In September, it announced medical cases had been viewed 1 billion times on the app.

Unfortunately, history is, again, not on it side. “They call it the Instagram for health,” said Matt Tindall, the general manager for consumer solutions at IMS Health, a health information and technology services company, “but we haven’t seen to date anything in health care scale like that.” Instagram has 400 million users, who upload more than 80 million images daily, according to the company blog.

In light of all these hurdles, Figure1’s best chance to reach “disruptor” status, then, might be to integrate with something like Watson, IBM’s Jeopardy!-crushing supercomputer, which already has access to de-identified medical records, experts told me. The computing giant also has at its disposal machine learning experts that would be able to glean valuable information from the real-world images Figure1 has collected. That’s the only way experts really saw it making big impacts in health care.

Unless that happens, Figure1’s most important use case might be in education. “Figure1 is a fantastic learning resource,” said Misra, the cardiology fellow. Doctors have always used each other to learn new things. “Figure1 scales that up.”

Through the app, he’s able to see how doctors in more than 110 countries deal with conditions like malaria, cholera, dengue—diseases he might never otherwise get to “experience” on his own. He’s able to join the conversation, ask questions, and bask in the collective knowledge of physicians around the world. Figure1 may not end up revolutionizing health care, but it could very well make its mark as a digital medical textbook that’s written not by a couple of experts sitting at an office desk, but instead by the nurses, physicians assistants, doctors, and specialists actively treating patients.

Daniela Hernandez is a senior writer at Fusion. She likes science, robots, pugs, and coffee.