Denther's truth extracting mechanism: how it works
As seen in our blog post “Over-treatments, an overlooked epidemic”, dentists sometimes have good reasons not to say the whole truth and nothing but the truth. With tough competition, rampant adverse selection, and terrible working conditions, you can almost understand them.
We are not saying that it happens every time, but 28% chance of over-treatment is too much in our eyes. This is why we have decided to find a way to get the truth every time. The requirements for such solution were clear:
- should be at least 10 times more reliable,
- it should create more value than harm to dentists and
- should be easy, fast and cheap.
With this in mind, we started testing out different ideas and quickly realised that controlling the information and incentives a dentist has is the key. This led us to the wonderful world of game theory (if you want to learn about game theory in a fun way, play this game).
Long story short, game theory helps us understand how people make decisions and how to effectively influence their decision making. If we could make dentist always decide to give the correct diagnosis, we would solve the problem, we thought.
Easier said than done. No dentist will ever publicly confront other dentist. We realised that dentists are biease by what other dentists think and it taints their decision making. So the first thing was to make sure dentists don’t know other dentists opinions. We expanded this rule to ensure that they don’t know anything about the case except the immediate relevant medical information. In Denther diagnosis, dentists solving the case can’t communicate with each other, or patient. This resembles a lot on game theoretic exercise called Prisoner's Dilemma, which is super smart and you should definitely check out this Youtube video.
However, this wasn’t enough. What we got by this was appalling diversity of opinions. Just like american journalist Ecenbarger and ETH Zurich researchers. Realisation was, dentists are still recommending treatments which they wanted to do, rather than giving the correct diagnosis. They did because it was the easiest thing to do, they were being paid to give their opinion and we had no way of proving they were wrong.
Luckily, we found a solution for this. Bear with us as we dive into details, this is the key to our solution. We told each dentist that we are asking 4 other dentists for opinion on the same case. But, none of them are getting paid anything unless at least 4 out of 5 say the same thing. If this does happen, these 4 will split the prize and get a reputation point. Reputation points are an objective, public score of how good a dentist someone is. It can really influence their career. On the other hand, the one dentist who said something different from the majority, gets no money, and loses 2 reputation points. Furthermore, if there is less than 4 agreeing, then none of them get money, and all lose 1 reputation point.
These rules make dentists think twice about each diagnosis. It’s easy to fail big - there are thousands of options, but only a handful of correct answers. Whenever a dentist thinks she sees something, she has to think if this is something that other dentists will identify? If so, she can’t miss it and if not, she can’t put it or she will lose.
Let’s get back to our requirements.
Is this 10 times more reliable?
To know this, we need a way to measure reliability. In a standard diagnosis, as we have seen, there is a 28% chance of wrong diagnosis. Second opinion probably has lower mistake chance, but not by high margin. Now, what is the chance of getting diagnoses a nonexistent dental pathology with Denther? There are on average 28 teeth in an adult person, and there are approximately 120 different dental pathologies. Let’s round that to 3000 possible findings. If an average person has 3 findings, this gives 30003, or 27 billion possibilities. Out of which only 3 are correct and have some indices. We won’t calculate it here, for the brevity of this post, but this already gives a hint that a chance that 4 out of 5 dentists agree on the same non-existent finding is way less than 1%. Let’s just say that out diagnosis is 28 more reliable than standard diagnosis when it comes to possible over-treatments.
But dentists among you might ask, what about the false negatives. What if our dentists miss a pathology. We admit, it is possible. Especially if examinations are not done correctly. However, the nature of dental pathologies is that they are never terminal. They will show up within the next 6 months and get treated, or not. Progression in 6 months will not result in anything drastic. However, we have standardised the examination procedure to ensure we capture all the most important stuff so there is no need for worry.
Does it create more value for dentists?
Economic theory teaches us that if a buyer doesn’t know the real value of the service, they will buy less of it. This is very true with dental opinion. Compared to dermatology, oncology or any other medical branch, it is generally easy to get free dental consultation. Patients know how unreliable it is, and refuse to pay for it. Denther diagnosis is proven to be correct and this makes its value clear to patients. Dentists who solve cases on Denther, thus earn money for giving their opinion and if they specialise, their revenue can be high enough to only do that. We actually aim at making it possible for our best dentists to earn more money this way than with clinical dentistry. Our projections show that money earned and benefits of objective reputation make up for money lost from unnecessary treatments. The byproduct is better health of patients.
Is it easy, fast and cheap?
This wouldn’t be possible if you had to meet each of 5 dentists in person. Luckily, dental imaging technology lets us record patients better than what a dentist can see and feel using only her senses and hands. We used this understanding to create a standardised examination procedure which results in a digitised records which enable remote, online diagnosis. Once you go to a dentist for an examination, that is it, everything else is done online with the speed of the internet. Last but not least is the price of a diagnosis. When we took into consideration desired revenues for dentists and our costs, we found that we are able to provide this service for 100€ which is what it costs to get a second opinion in the US and Western Europe.