Digital health · Crowdsourced diagnosis

CrowdMed

A marketplace where patients paid a monthly fee to post an undiagnosed case, and a crowd of self selected “medical detectives” ranked the possible answers. Top tier venture money, one independent peer reviewed evaluation, and years of published objection from physicians.

One hundred strangers

In 2003 Carly Heyman was eighteen. She fell into a deep depression, slept fourteen hours a day, gained fifty pounds and woke in cold sweats. Her parents took her to sixteen specialists and ran up more than $100,000 in bills. None found the cause.

The answer came from a team at the NIH Undiagnosed Diseases Program: fragile X associated primary ovarian insufficiency, which affects about one woman in fifteen thousand. A hormone patch cleared her symptoms.

Her older brother Jared had built a career on survey methods and prediction markets. He ran an experiment. “With an early prototype, we tried to retroactively crowdsource the answer to my sister’s case using 100 randomly selected people and showing them the exact same information that her doctors had before she was diagnosed,” he told Smithsonian in 2015. The crowd named it in days. That result became a company.

The gap it aimed at

Formal help for hard cases was narrow. The NIH Undiagnosed Diseases Network screened applicants and required a referral. Best Doctors served only employee beneficiaries. The Cleveland Clinic ran a records based second opinion with no patient contact. The National Organization for Rare Disorders told NBC News that rare disease patients wait an average of seven years for an accurate diagnosis. CrowdMed offered no referral, no eligibility screen, and a conversation.

Three founders and a batch

Jared Heyman had founded the survey company Infosurv in 1998 and grown it past $20 million in aggregate revenue. Axel Setyanto, previously at Loku, was lead developer; Jessica Greenwalt, founder of the design firm Pixelkeet, was lead designer. CrowdMed came out of Y Combinator’s Winter 2013 class and launched its public beta at TEDMED that April with $1.1 million from NEA, Andreessen Horowitz, Greylock Partners, SV Angel and Y Combinator. By May 2014 the round stood at $2.4 million, adding Khosla Ventures and the actor Patrick Dempsey.

What a case cost

A patient filled in a questionnaire, uploaded records, and the case went live. Solvers suggested diagnoses and bet points on them. An algorithm ranked the suggestions, a licensed physician moderated, and the patient got a report to take to a doctor.

The price moved. Popular Science reported a $50 posting fee and a $99 monthly package in 2015; USA Today put it at $150 to $350 that November; by late 2016 The DO reported $149 to $749 per month, still the range Peninsula Press found in 2020. Patients also set aside rewards of $200 to $1,000 for the most useful solvers, and at least half of what they paid went to the detectives.

Who those detectives were is what drew fire. Anyone could sign up. In the 2016 evaluation, 58.3% of active solvers worked or studied in the medical industry, including 36 physicians and 56 medical students out of 357 people. Danyell Jones, then chief executive, called the mix deliberate, arguing that US physicians share a training background and therefore share blind spots.

Two papers and a rebuttal

In January 2016 the Journal of Medical Internet Research published an independent evaluation by Ashley N. D. Meyer, Christopher A. Longhurst and Hardeep Singh. CrowdMed supplied the raw data and, the paper states, did not fund, commission or shape it. Across 397 completed cases, 50.9% of patients said they would recommend the service and 59.6% said it gave insights bringing them closer to a correct diagnosis. No eventual diagnosis was verified. The authors also flagged “the potential legal ramifications of giving individuals without medical degrees (who make up a substantial portion of the case solvers) the ability to render diagnostic opinions”.

Five months later the same journal published a second paper, on health care use. Visits fell from 1.07 to 0.65 per month and charges from $719.70 to $516.79. Its funnel was stark: 546 eligible users, 45 completed surveys, 13 in the final sample. The disclosure reads that “Evidation Health, Inc. was paid by CrowdMed, Inc. to conduct this study.”

The lawyers moved first. In April 2015 Stanford Law School’s Law and Biosciences Blog set out California prohibitions on unlicensed diagnosis and on the corporate practice of medicine and asked, “But one obvious question is: How is this not the unauthorized practice of medicine?” No enforcement action has been found in the public record.

The sharpest clinical criticism came from Jeff Russ, then a child neurology resident at UCSF, on KevinMD in March 2017 and in the San Francisco Chronicle that May. His argument turned on sequence: a crowd ranks by popularity, a physician ranks by danger. He cited a case aired on the podcast Reply All in which imaging for a possible arterial dissection sat on the crowd’s list below chiropractic manipulation, which the patient pursued.

The last numbers on file

The 2016 crowdfunding raise put the books on the record: revenue of $282,981 in 2015 against a net loss of $802,014, six employees. Then the filings thin. The last, covering 2019, lists one employee, revenue of $127,787, cash of $79,073 and long term debt of $4,671,564. Jones signed as chief executive, Lee Jared Heyman as chairman. Nothing has been filed since June 2020.

What is proven, and what is still claimed

Read plainly: the strongest independent finding is that most surveyed patients said the experience helped, and the authors of that finding wrote in the same paper that nobody had checked whether it did. The company’s 70 percent and the researchers’ 59.6 percent do not measure the same thing, and only one was produced outside the company. The criticism was never that crowds cannot diagnose; a 2024 comparison in BMJ Quality and Safety found groups beat individuals. It was that a popularity ranking is a poor instrument for ruling out what kills quickly, and that patients who had run out of options were paying for advice they had no way to grade.

What to watch

• Whether anyone publishes downstream outcome data. Both 2016 papers named that gap; neither closed it.
• Whether the acquisition stated in a speaker biography is ever documented anywhere else.
• Whether the notes sold to 608 crowd investors in 2017 were converted, repaid or written off.
• Whether crowdmed.com resumes accepting cases, and under whose oversight.

In their words

Related companies

• Alfie: Also in Healthcare Services. A New York telehealth clinic that used an AI model called ObesityRx to match patients with weight loss medications and behavior changes
• Experiment: Also in Healthcare Services. A crowdfunding site where a graduate student can raise three thousand dollars in a month instead of waiting a year for a grant
• Science Exchange: Also in Healthcare Services. A marketplace for ordering experiments that became procurement software for big pharma, and that sourced the laboratories for the largest public test of whether
• FidoCure: Also in Healthcare Services. A tumor sequencing service that routes dogs with cancer toward human targeted therapy drugs, compounded to canine doses

Sources

1. Public recordCrowdsourcing Diagnosis for Patients With Undiagnosed Illnesses: An Evaluation of CrowdMedJ Med Internet Res 18(1):e12 · Jan 14, 2016
2. Public recordThe Impact of an Online Crowdsourcing Diagnostic Tool on Health Care UtilizationJ Med Internet Res, PMID 27251384 · Jun 2016
3. Public recordCrowdsourcing a diagnosis? Accuracy of the size and type of group diagnosisBMJ Qual Saf 33(12) · Nov 20, 2024
4. Public recordMapping of Crowdsourcing in Health: Systematic ReviewJ Med Internet Res 20(5):e187 · May 15, 2018
5. Public recordCommentary on the CrowdMed evaluationAHRQ PSNet · 2016
6. Public recordSEC EDGAR filings, CrowdMed, Inc., CIK 0001687611SEC · Oct 2016 to Jun 2020
7. IndependentThe dangers of inexpert diagnosis from a noisy crowdKevinMD.com, Jeff Russ · Mar 18, 2017
8. IndependentThe dangers of crowdsourced medicineSan Francisco Chronicle, Jeff Russ · May 2017
9. IndependentCrowdsourcing Medicine?Law and Biosciences Blog, Stanford Law School · Apr 23, 2015
10. IndependentIn the Wild West of online medicine, crowdsourcing is the next frontierPeninsula Press, Jasmine Kerber · Jan 6, 2020
11. IndependentHow internet detectives are helping diagnose rare disordersNBC News, Christina Couch · Mar 6, 2017
12. IndependentWould you trust Internet sleuths to diagnose rare disease?USA TODAY, Christina Farr · Nov 30, 2015
13. IndependentYou Can Now Crowdsource Your Medical Diagnoses, But Should You?Popular Science, Alexandra Ossola · Jul 7, 2015
14. IndependentCrowdMed’s ‘medical detectives’ tackle tricky cases onlineThe DO, Laura Selby · Nov 21, 2016
15. IndependentCrowdMed and crowdsourcing of medical advice onlineiMedicalApps, Brian Chau, MD · Mar 25, 2016
16. IndependentYC-Backed CrowdMed Launches To Crowdsource Medical DiagnosesTechCrunch, Ryan Lawler · Apr 16, 2013
17. IndependentPatrick Dempsey contributes to CrowdMed’s $2.4M seed roundMobiHealthNews, Aditi Pai · May 21, 2014
18. CompanyCan You Crack a Medical Mystery? Interview with Jared HeymanSmithsonian Magazine, Li Zhou · Jun 11, 2015
19. CompanyCrowdMed Wefunder profile and quick factsWefunder · accessed Sep 22, 2026
20. CompanyJared Heyman speaker biographyTEDMED · accessed Sep 22, 2026