Can you crowdsource medical diagnosis?

crowdsourcing-2Crowdsourcing has never been more popular.  My last blog showcased the latest NASA Hackathon that saw thousands come together to offer solutions to scientific problems.  The wisdom of crowds was also incredibly prominent during the aftermath of the Boston bombing, with people from around the world looking over photos from the event in an attempt to track down the bombers.

Wonderful though these things are, would you want to trust the crowd with your health?  New start-up CrowdMed aims to find out.

The site, which was launched last week, aims to apply the wisdom of crowds to our medical problems.  The concept is a fascinating one.  You don’t have to be medically trained to join the site, with the only criteria being a willingness to investigate cases.

As with many other crowd based services, users will be rated and given points through their use of the site.  These points act as a sort of virtual currency that can be bet on the correct diagnosis from a list of suggestions.  These bets form the basis of a prediction market, with the suggested diagnosis rising and falling in price, just like stocks on the stock exchange.  The site then uses algorithms to calculate the probability that each diagnosis is correct.

Once the crowd have arrived at their consensus, the CrowdMed site shows patients the top three suggestions, which they are then free to explore with their doctor.  If the diagnosis proves correct, the doctor is encouraged to feed the results back into the site, with the participants rewarded accordingly.

The results thus far look promising.  Participants have ‘diagnosed’ 20 test cases with a good degree of accuracy.  As numbers increase from the 700 or so people currently using the site, the owners expect accuracy to increase.

Whilst there will hopefully be mainstream applications for the site eventually, it’s initial aim is to target around 7,000 rare diseases.  These alone however have around 30 million victims in Europe alone, with some 40% of these people either going undiagnosed or having a misdiagnosis.

Crowdsourcing meets big data

Could the future see heavy duty computing such as IBM’s Watson teaming up with the crowdsourcing of CrowdMed?  It’s something IBM are particularly keen on.  Claudia Perlich, who helped develop Watson, imagines it teaming up with CrowdMed to work on tough diagnoses.

“If Watson could get hold of what people submit to CrowdMed, I would love it,” she says. “I absolutely agree with the premise that the big problem of the medical system is that we don’t have sufficient information sharing.”


11 thoughts on “Can you crowdsource medical diagnosis?

    • As with most things, I suspect the first people that will use this kind of system are those with nothing to lose. Once data from that comes in it will inform the mainstream about the viability of it I suspect.

  1. I think this is a great idea. So much evidence out there that innovations often come from people outside the field of expertise. Why not in healthcare?

  2. I love this idea. With all the info that is out there on medical conditions, I think that the crowds that would attempt to guess these conditions would be the well informed crowd and what with Watson getting into the action, I think that the results would be worth looking into esp. with the amount of info at his disposal. What an innovative idea! Great idea for a post. I hadn't heard of crowdsourcing a medical diagnosis before!

    • And the nice thing is that with the amateur diagnosis being aggregated, it tends to weed out the poorly informed attempts, so the theory goes that you 'should' get as good, if not better insight than you would get from a professional.

  3. Love it! I am a firm believer in tapping the wisdom of the crowds, yet we do so little of it. For medical purposes? Why not? Yes, there are some serious issues at stake here, but if the process goes as explained, with the ultimate diagnosis or decision-making residing into doctor's hands, what's there to lose?

    • It's interesting in relation to your latest post on innovation I think. The book on wisdom of crowds by James Surowiecki was published almost 10 years ago, but it's only really now that we're seeing a whole lot of crowdsourced applications coming to market.

  4. Would be curious to know how the diagnoses were confirmed. I suspect this could easily turn into an extension of the familiar patient pastime of doctor-shopping for an acceptable diagnosis; an acceptable diagnosis is usually available somewhere in return for a sufficient fee.

    • The phenomenon of shopping for an "acceptable" diagnosis is rampant in psychiatric situations. For example, many patients would insist on being diagnosed with an Axis I condition and are "disappointed" when told their presenting complaints are more congruent with an Axis II diagnosis. As you said, for a sufficient fee, they will eventually get the label they desire of some "trendy" illness, with a prescription to treat it no less.
      While "crowd" is a place for searching when a rare illness keeps being missed by a primary care physician (I think in UK, they are called GP), caution definitely needs be exercised. Consumers don't always know best.

  5. It should be no surprise that doctors aren't particularly good at diagnosing unusual diseases/health problems. The training of doctors is rote and thinking is generally frowned upon. After qualifying, doctors get most of their new information from the pharmaceuticals companies. Few have the time or interest necessary to keep up with research and new information. Conversely scientifically minded individuals lacking these constraints may be very good at analytical reasoning and may also be very interested in the latest clinical and research data.

    Our approach to health service provision is woefully out of date, reflecting its roots in pre-Victorian practices. It is more than past time that we restructured our entire health delivery system (including training of doctors) to move into the 21st century rather than remain in the 19th.

Leave a Reply

Your email address will not be published. Required fields are marked *