Wisdom of Crowds for Rare Diseases: CrowdMed

Traditionally, access to the wisdom of crowds in healthcare has been limited to doctors: hurried hospital rounds, case studies during medical school, or discussions with an interdisciplinary care team during residency. A new startup, CrowdMed, seeks to bring the wisdom of crowds to the patient by helping to solve the world’s most difficult, rare, and complex chronic medical cases through online communication, collaboration, and brainstorming.

The inspiration for CrowdMed came from Jared Heyman, who saw his sister battle an undiagnosed rare illness, traveling to numerous specialists, and spending hundreds of thousands of dollars with no solution. She finally found a diagnosis after three years. After launching CrowdMed, Heyman posted his sister’s case on the site out of curiosity. The correct diagnosis was found in three days.

Harnessing the power of the crowd to discuss and brainstorm root causes of illnesses has tremendous potential to disrupt the healthcare industry. Currently, healthcare issues are filtered through a primary care physician, who has no way of keeping up with the ever-evolving literature on the 7,000 rare diseases known to mankind today. Referrals often result in specialists viewing the patient’s symptoms from the narrow lens of their own specialty, without holistically understanding their case and complex interplays between specialties. The reality is that no one single doctor can remain abreast of the vast amount of medical knowledge today. Although we would all like to believe we could be diagnosed properly if we just found that one genius doctor, the truth is that Dr. House doesn’t really exist. Enter the crowd.

How it works:

Pathways to a Just Digital Future

Watch this tech inequality series featuring scholars, practitioners, & activists

• A patient with an unsolved medical mystery posts a case online by answering a comprehensive set of medical questions and uploading relevant images and diagnostic lab results.
• CrowdMed’s community of 15,000 registered “medical detectives” collaborate on the case using discussions, chat rooms, collaboration, and brainstorming, ultimately voting on what they think the most likely diagnosis is. Medical detectives are vetted based on a performance-based reputation system – similar to TopCoder and Upwork, one’s CrowdMed reputation is built up by making accurate suggestions or voting on those suggestions made by others. As ranking increase, medical detectives earn the right to work on more advanced cases.
• A patented prediction market technology accurately identifies the most probable diagnoses and solutions based on, amongst other factors, the historical accuracy of the medical detectives and the number of points bid on each diagnosis.
• This algorithm calculates the top three most probable diagnoses and sends them to the patient, who can print them out and bring them to his/her doctor for further exploration.
• Patients are charged $50 as a deposit and a subscription fee to post cases which can range from $99 to $249 depending on the number of medical detectives desired and presence of expert case review moderators. Patients can also include small compensation rewards to attract the best case solvers.

Benefits:

• Huge savings. Dramatic potential for time and cost savings to the healthcare system and to patients themselves. The average CrowdMed patient has been suffering symptoms for over eight years and has spent tens of thousands of dollars on seeking a diagnosis, not to mention countless ER visits, hospital stays, medication, and other costs. Cases are solved on CrowdMed in anywhere to a few days to a few months, with an average case resolution time of 2 months and average cost of $500. Heyman’s target market is the 2-3% of consumers that are responsible for 12-15% of the total healthcare costs in the U.S. Crowdsourcing a diagnosis would be exponentially cheaper and faster, benefitting both the patient and the healthcare system.
• Results. Despite its nascency, CrowdMed has already resolved over 1,000 complex medical cases, with a 70% success rate. 95% of diagnoses have only arisen once on the site, highlighting how rare these diagnoses are. Resources like WebMD exists for routine symptoms, but few exist for chronic long-term cases.
• Reducing information asymmetry. By bringing crowd wisdom to patients, they are now privy to medical knowledge such as case consults and medical journals that were only available to doctors, encouraging patients to be more invested in their own health.

Challenges:

• Not all medical detectives are licensed doctors. Some are patients with their own experience with rare diseases, others might be kooks. The strength of this model is that it rewards performance instead of formal background, which is the opposite of how the system currently works. CrowdMed’s top performers don’t necessarily have the highest formal medical credentials, which demonstrates the limitations to relying on the “best” specialist in a field to diagnose you. However, a challenge arises if a wrong diagnosis is discovered, which leads to extra time the physician has to spend convincing the patient that the diagnosis isn’t correct.
• Some of the best voices may not be heard. Crowdsourcing tends to attract people with time on their hands or a younger generation of medical detectives, meaning that some experts may not be present because of time constraints or technology aversion.
• Crowdsourcing may pose constraints when trying to root out the cause of a complex medical illness instead of just surveying people for their opinion. Different people describe the same symptoms in dramatically different ways, and it can be a challenge for doctors to truly understand complicated symptoms without examining the patient firsthand.
• Patient safety could also be a concern – although identifying information is hidden, most of these conditions are rare and it might not be too hard to find out who a patient is if one really tried. Healthcare also remains deeply personal, and it might be challenging to overcome this issue to post detailed medical information online to non-licensed personnel.
• As to be expected, many physicians are extremely resistant to this idea. It’s challenging to get doctors to admit they don’t know everything, collaborate with people they don’t know, put aside their egos to brainstorm diagnoses that may ultimately be wrong, and admit that a non-credentialed person may be better informed than them. Was that a bit harsh? Let me rephrase. Healthcare in the United States is intrinsically an individualistic practice, and changing culture can be challenging.

Harnessing collective intelligence to crowdsource diagnoses for the most expensive and difficult patient cases poses massive disruption potential to the healthcare industry and to the role of the doctor as we know it. CrowdMed plans to partner with health insurers and providers to scale nationally, aiming to save billions of dollars in unnecessary healthcare costs and prevent needless suffering for patients.

The number of success stories from CrowdMed is countless. The growth potential is huge if CrowdMed can overcome resistance from physicians and bioethicists and leverage society’s growing cultural and psychological comfort with crowdsourcing their most personal issues.