Getting the right diagnosis is harder than you think. (it seems docs aren't aware of how hard it is, either). Doctor errors in diagnosis are surprisingly high. And a bigger concern is that doctors are usually not aware of the difficulty and errors). So patient frustration with doctors comes with good reason. Artificial intelligence is now becoming available to help. But so far, docs have been slow to adopt diagnostic software. The study cited below in the British Medical Journal shows the good reason: previous software got the right diagnosis only 35% of the time. What the docs failed to realize, though, is that doctors are worse. Yes, docs get easy cases right about 55% of the time. But they correctly diagnose more challenging cases less than 10% of the time. And again, they usually think they're right. Therefore, they don't know that they need to keep checking. How can we develop a computer that beats the world chess masters easily - but we can't develop one that helps your typical doc? It has to do with the process. To date, the software has been developed using "expert logic". This means rules are obtained for experts and programmed into the software. But Toyota gained prominence through a continuous improvement process. And Deep Blue became Very Deep Blue through a continuous learning process. If IBM can make Deep Blue continue to learn, why can't the diagnostic software developers? There is one out there now doing just that - K Health. I've met this team and plan to do some work with them. Will you help me by trying it? Download the app. Take it for a spin. And comment below. https://www.bmj.com/content/351/bmj.h3480 https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/1731967
For more information, contact us at 859-721-1414 or myhealth@prevmedheartrisk.com. Also, check out the following resources: