When I was in training, we discussed difficult cases in Resident’s Report or at Grand Rounds. Once I was out in practice, I got help by looking stuff up in books (remember them?) or calling in a consultant or two. For some cases, the really hard ones, this is simply not enough brain power to get to the correct diagnosis quickly. For rare diseases or unusual presentations of common diseases, workups can be difficult and protracted.
The Founder’s story
Jared Heyman knows this all too well. His sister went from a young healthy teen to someone who had to drop out of college because of depression and a fatigue so profound she could only watch TV or play solitaire. She would also wake up several times a night with nightmares and cold sweats. Eventually, she became suicidal.
Over the course of three years, she saw sixteen different medical specialists (all seen individually) racked up $100,000 in medical bills but to no avail. They couldn’t figure out what was going on. She had become something you really don’t ever want to be in medicine: the proverbial interesting case.
Eventually, the doctors were able to make a diagnosis. She had a rare disease that only affects 15,000 females. She was treated and is doing fine now, but her story inspired her brother to do something to help other people like her.
Jared Heyman founded CrowdMed to crowdsource medical diagnoses that had stumped the doctors of people with difficult to diagnose conditions. That’s right, he designed a platform to bring together people that he calls Medical Detectives—people who want to help solve the puzzle. I had a chance to interview Jared on TDWI on the Radio this week. In preparation for the interview, I signed up to be a Medical Detective myself.
Many of the Medical Detectives are retired allopathic physicians, but some are in active practice and some are naturopaths. You don’t have to be a physician to be a Medical Detective. You just have to be motivated to help solve the patient’s problem. Once you sign up, you are oriented to the site by doing a few test cases. The first one I tried out was so complicated that I had to Google the symptoms and look up some of the more arcane tests that he had had performed. I also read all the questions and answers from the other Medical Detectives as well as the patients. I, then, came up with a couple of diagnoses, mainly by picking ones other detectives had chosen for what they described as compelling reasons. The last step in the process is to assign a probability to them by assigning the diagnoses suggested by the Medical Detectives on the cae some of the points I have to make diagnostic bets with.
Jared told me that CrowdMed’s “patented prediction market system” collects the bets and then develops a list of the most likely diagnoses and solutions. He doesn’t have any formal pilots or studies to validate his outcomes yet, but he said that the feedback they have from patients who have participated is that the crowdsourced diagnoses were correct 80% of the time.
Crowdsourcing other complex problems
It will be very interesting to see how this plays out. There are instances of application of crowdsourcing to solve complex problems in other fields. One particularly fascinating example is the gaming software site, Foldit (whose tagline is “Solve Puzzles for Science”). Nature Outlook reports on the use of Foldit to solve a protein folding problem. Here is how they described it:
“Following the failure of a wide range of attempts to solve the crystal structure of M-PMV retroviral protease by molecular replacement, we challenged players of the protein folding game Foldit to produce accurate models of the protein. Remarkably, Foldit players were able to generate models of sufficient quality for successful molecular replacement and subsequent structure determination. The refined structure provides new insights for the design of antiretroviral drugs.”
It makes intuitive sense that combining the different perspectives of individuals with varied backgrounds and piling on the brainpower of the crowd has the potential to solve previously unsolvable problems, including difficult diagnoses. I look forward to seeing a formal evaluation of the outcomes of this new approach.
Here is a link to my podcast interview with Jared Heyman, CLICK HERE.