This may seem like a silly question. How is it possible that anyone but experts can diagnose us? We are an expert-driven culture, with a reliance on further and further specialization. It’s generally assumed, for good reason, that one person who’s very knowledgeable is better than a group of people who aren’t as knowledgeable.
Not only that, especially in medicine, it seems that any attempt to rely on people outside of the world of “expertise” (especially on the Internet) can not only end in frustration (think of trying to ask your Facebook friends to diagnose you) but, in the case of things like the anti-vaccination movement, it can be downright dangerous.
So, back to the question at hand: why on earth would anyone trust anyone but individual experts to give them medical advice?
CrowdMed is horrible
With all the above questions one might ask: why on earth should a site like ours, CrowdMed, exist? CrowdMed uses “crowdsourcing”, the word that essentially describes the opposite of an expert-based system: we use lots of people to help diagnose medical conditions instead of just one or a few.
Here are the problems you could cite (and many do):
- We don’t have any requirements for who can contribute to our site. From nurses to former patients to people who just like trying to solve mysteries, our site is open to all “Medical Detectives”, as we call them.
- Formal medical credentials make absolutely no difference to our system. A doctor with 50 years of experience has the same voice (at least initially) as a punk 20 year old kid who spends his free time arguing on reddit.
- There’s no real-life accountability. All contributions by our Medical Detectives are anonymous on our site, in order to, among other things, protect them from liability. Someone could be an absolutely horrible Medical Detective and still be a part of our site.
Viewed through the lens of an expert-based system, we are horrible. We don’t do any of the things that expert-based systems rely upon, including licensing, valuing formal credentials, and real-life liability.
CrowdMed is amazing
But here’s the weird thing: CrowdMed has helped solve hundreds of the hardest medical cases in the world. People who have spent decades going to individual experts in vain have come to our site and been successfully diagnosed in as little as two weeks.
See our last post for more details about the success we’ve achieved, including our 55%+ success rate.
So, clearly CrowdMed is only horrible when viewed in the context of the expert-based paradigm.
Why CrowdMed is different than Facebook
So if crowds are smarter than individual experts, why not just post medical problems on Facebook? What makes CrowdMed any different?
Our system. CrowdMed isn’t just a place for people to post medical problems and get a bunch of randoms to submit answers. It’s a system that actively filters, weighs and ranks diagnostic suggestions, solution suggestions, and the Medical Detectives themselves.
In other words, CrowdMed doesn’t care about degrees, it cares about results. While anyone is allowed to join, our system gives much more weight to the Medical Detectives who have already proven themselves to be successful on our site.
Each one of our Medical Detectives has a “DetectiveRating” (based upon the PageRank system Google uses) that rewards successful detectives. Detectives are ranked based on how well they diagnose, their interactions with the patients, and more. In the end, the best detectives are given more influence in our system.
More non-experts = a better system
That may all be nice and true, but why let people that are not even medically trained into our system? Why let nurses, who technically are not allowed to diagnose patients, contribute? Why let medical students contribute? Since when is it okay for pharmacists to diagnose?
And this is where crowdsourcing can really upend the way we look at medicine and the expert model in general.
Let’s look at our data.
As we mentioned in our last blog post, over the life of CrowdMed, we’ve had a 55% success rate with all our cases.
But here’s some more data that shows the power of diversity:
In August, we had a sudden jump in our success rate. It shot up from 55% to 60%. For the entire life of our company, we had stayed steady at that 55%, so we were intrigued as to why this would be.
We hadn’t made any dramatic changes in our system in this time. Not much really had changed except one thing: the diversity of our Medical Detectives.
Here was the makeup of our Medical Detective community before August:
As you can see, the vast majority of our community at this time was made up of medical students and medical doctors. We had a few nurses, researchers, and “others” (medical professionals that didn’t fit into neat categories). But the vast majority held positions that qualified them to diagnose in the present or the future.
And here are our active Medical Detectives from August to the present:
As the chart shows, the percentage of nurses using our system shot up. So did “other” and even “alternative medical professionals”.
In other words: diversity increased the chances of CrowdMed helping patients find a diagnosis.
In other other words: having “non-experts” use our system doesn’t hurt our system. In fact, it increases our chance of success.
Having a narrow profession contribute to our system may help in some areas, but in others it may not. This is the value of a good crowdsourcing system: the ability to take what is special and unique about each person and apply it to a larger goal.
Can non-experts do better than experts?
Okay, so diversity helps. But maybe reducing medical students is what helped us? How do we know for sure that non-experts actually help our system? Maybe medical students just hurt our system?
To answer this question, we decided to go deeper into our stats. It was fine to look at CrowdMed improving as a whole, but we needed statistics about specific professions to really make this case.
What we found surprised even us.
We took the points each profession “bet” on a diagnosis, and how much they “won”. As mentioned above, our point system is how we track the success of our Medical Detectives. By tracking the amount of points won per case by each profession, we were essentially measuring their “accuracy” in the same way a video game would.
Here’s were the results:
Rather than medical students being the problem, it seems that they are, in fact, some of our most valuable contributors. As a group, they are the most accurate in their diagnosis suggestions.
Nurses followed close behind, and so it is no wonder that they have helped our community.
But, of course, the most surprising result is that physicians rated below both groups. It is perhaps too early to draw broad conclusions from this.
But one conclusion we can definitely draw: the idea that non-experts will bring down our system is clearly incorrect, and the theory that they help our system clearly is correct.
We can now see why the very criticisms made against us that we listed above are, in fact, the key to our success:
- Reducing barriers to entry increases the diversity of our medical detectives, which in turn leads to a better system with better results.
- A system that allows anyone to enter needs a filter or else it becomes just like Facebook. That’s why we rely on our Medical Detectives’ results to rank them. By judging them on how they do, we can keep our system open and diverse, thus resulting in more accurate diagnostic suggestions.
- While there isn’t “real-life” accountability, our system has a built-in measurement system that allows us to filter out the people who would hurt our patients through their suggestions. This allows us to provide our Medical Detectives with a feeling of freedom while still giving high quality results.
We are so excited to share these results because they are a sign that as CrowdMed grows, and, in turn, its Medical Detective community grows, the more successful our platform will be.
In other words, the bigger CrowdMed gets, the more effective we will become.
To sum up: this is great, but this is just the beginning. We are so moved to have helped over 500 patients, but we see it as the tip of the iceberg, and the beginning of something great.
And we are so excited to have you on board for the ride.
If you are struggling with finding a diagnosis, click here to submit your case to CrowdMed. Our community of “Medical Detectives” will do their best to help you get answers.
Every person who contributes to our community helps these numbers go up! Join our Medical Detective community and make a difference in the lives of people desperate for a diagnosis. Find out how here.
Featured image source: Flickr