In the Wild West of online medicine, crowdsourcing is the next frontier

 

Dominik Sheehamski had been perfectly healthy until five years ago. But suddenly, the then-32-year-old began to experience ringing in his ears, vertigo, stomach bloating and odd, fluctuating sensations of pressure in his head and body. When specialist after specialist failed to diagnose and treat his mysterious illness, he turned to the internet.

Sheehamski had heard of the new Netflix show “Diagnosis,” in which Dr. Lisa Sanders helps patients with rare conditions “crowdsource” their diagnoses. Dr. Sanders is a professor at the Yale University School of Medicine, and she’s been writing about medical mysteries in a New York Times column since 2002. Sheehamski, a New York-based scientific grant writer with a Ph.D. in neuroscience, hoped she might take his case.

“I thought, ‘why not?’” said Sheehamski, who sought advice using a crowdsourcing Facebook group but did not try CrowdMed. He’d already seen multiple doctors and combed the internet for information. What harm could one more attempt to find a cure do?

While patients commonly Google their symptoms early on, crowdsourcing diagnoses has gained traction in recent years as an unconventional approach to solving particularly tricky medical mysteries—especially when doctors struggle to resolve an illness. Sanders’ show reflects the popularity of the trend.

CrowdMed, a San Francisco-based startup founded in 2013, offers a subscription-based website for crowdsourcing medical diagnoses. When people join CrowdMed — the subscriptions range from $149 to $749 per month — they answer a health history questionnaire and upload test results and relevant medical records. Then, the company matches each subscriber with a “medical detective.” Some, but not all, of CrowdMed’s approximately 10,000 “detectives” are licensed physicians; the company’s website says 60 percent of them “work in or study medicine.”

“That’s actually very intentional,” said CrowdMed CEO Danyell Jones, who acknowledges that the choice “sounds counterintuitive.” Jones said the company prefers having medical detectives with a variety of educational backgrounds because U.S. physicians all receive similar training, so “if there is an error in them being able to catch something, that’s an error that’s going to be duplicated.” CrowdMed’s detectives include “licensed physicians, medical students, physician assistants, chiropractors, scientists, naturopaths, and health-care aficionados,” according to the company’s website.

At least half of the fees patients pay to CrowdMed go to medical detectives, Jones said. Detectives who help solve cases earn cash rewards.

CrowdMed, which does not have any direct competitors in the medical crowdsourcing space, has attracted funding from prominent investors including the Emerson Collective’s Laurene Powell Jobs and 23andMe founder Anne Wojcicki. A 2016 study in the Journal of Medical Internet Research examined 397 CrowdMed cases and found that about 60 percent of patients “reported insights that led them closer to the correct diagnoses.”

Stanford Law School Professor Hank Greely, who specializes in the ethics and legality of new biomedical technologies, said that although collecting more information could help solve medical mysteries, “The problem is you need a really good curator, somebody who really can distinguish between good ideas, plausible but less certain ideas, clearly bad ideas and completely loony ideas—because my guess is you’ll get all of them.”

CrowdMed aims to reduce the influence of loony suggestions by appointing a moderator, who must be a licensed physician, to oversee each discussion. Jones also emphasizes the role of CrowdMed’s patented algorithm, which she likens to a futures market, a type of economic exchange in which people decide to buy and sell commodities at a later date.

At least a dozen CrowdMed users per case bet on which suggestion they consider the most likely diagnosis, eventually producing a list of three “most likely” outcomes. “It’s not just a simple weighing of everything,” Jones said. “The algorithm that we have is what makes us different.”

In the internet age, patients regularly try to make sense of their symptoms using WebMD or other online resources, including discussion forums.

“I see people coming in with such misinformed opinions based on online research,” said Dr. Arielle Levitan, an internist in Highland Park, Illinois. Levitan said many patients develop theories about problems they believe they have based on Google searches and stories they’ve read in online chat rooms.

Although she said there “may be a role” for internet information and discussion forums, Levitan noted, “There’s a reason we go to school for a really long time and spend many decades seeing patients and getting experience. It’s really hard to replace that.”

Unlike WebMD, which many patients check before visiting a doctor, CrowdMed comes in later in the diagnostic process. Many of those who crowdsource their diagnoses have already turned to doctors, and although medical and legal experts say there could be a place for crowdsourcing tricky cases, they express some hesitations about how to keep crowdsourcing safe.

Sheehamski’s email to Dr. Sanders returned an auto reply saying the doctor was very busy but that a Facebook group associated with the show might be a good resource. Given his limited options, he posted several paragraphs about his medical history on the page and asked if anyone could suggest a diagnosis. Ideas ranged from food allergies to high blood pressure to a pancreatic tumor.

Because of his background in science, Sheehamski trusts himself to rule out wacky suggestions.

“Anything that was proposed, I’ve already done research on in the past,” said Sheehamski, who has taken many medical tests over the past five years and does not believe he has any of the conditions group members suggested. The group is made up of “just regular people,” he noted.

“I’ve given up,” Sheehamski said. “I don’t think anyone has a cure for me. I’ve been to the finest specialists in just about every area.”

He said the most plausible suggestion he has received from conventional doctors is that a stomach problem is putting pressure on his vagus nerve. The vagus nerve stretches from the head to the digestive system and communicates sensory impulses, which could link Sheemhanski’s seemingly disparate symptoms.

Sheehamski notes that he’s not entirely disheartened, though. He said CrowdMed sounded interesting, but he’s not sure whether he’d try it. A combination of two medications that doctors prescribed to him before he joined the Diagnosis show’s Facebook group has brought his symptoms down to a more manageable level.

“It’s not a cure, but it’s a pretty good Band-Aid,” he said.

Many of Sheehamski’s fellow Facebook commenters write that they have a friend or family member with similar symptoms, but a few comments make little sense. Under one mother’s description of her son’s medical problems, including autism, allergies and an unusually high heart rate, someone has suggested that the child consume juniper berry, sunflower oil, hawthorn berry, dandelion root, sasparilla root and seabuckthorn.

“I think it would be really dangerous to try these given the history of deadly allergies,” the mother responded.

Patricia Zettler, a law professor at Ohio State University, said a patient could bring a claim against a crowdsourcing company, but the success of any hypothetical lawsuits would depend in part on exactly what service such a company offers. Are they practicing medicine, which would open the opportunity for malpractice lawsuits, or offering some other resource?

A December 2019 search of case dockets in Unicourt, an online legal database that includes records of civil lawsuits filed in San Francisco, did not return any record of suits against CrowdMed.

“We are not practicing medicine,” Jones said. “We’re a diagnostic aid.”

She added that the company’s services are intended for “people with unresolved medical conditions,” most of whom have already been to the doctor in person.