With medicine, you really don't have time to go Google something. Even in a slow-paced assessment, you need to gather data quickly and effectively, and make decisions. Patients might in the future be happy filling out long series of probing questions and then waiting for a few hours/days while someone does some Googling, but I doubt it. The practice of medicine by human Doctors probably can't be much further optimized from what it is right now.
The practice of medicine by human Doctors probably can't be much further optimized from what it is right now.
Maybe this is true for the 95-99% most common of conditions. Doctors can be shockingly bad at diagnosing anything they don't see on a daily basis. (rare-ish conditions, unusual presentations, or anything discovered since they studied medicine)
Google search is probably very inefficient for this kind of thing, but specialised search engines based on bayesian networks or similar might get you some leads on what to test for a lot faster than going from doctor to doctor in the hope that one of them might have an idea that pans out.
We had AI systems that outperformed doctors in limited domains as early as in 1972 [1].
We've of course had massive improvements in machine learning in the 43 years since then. I think the practice of human doctors could be massively improved by having them stop spending time on things that machines are much better at. Hopefully things like IBMs Watson can make a difference in this regard.
An intelligent database that can cross link a million conditions and varieties with voice-recognized list of identified symptoms?
I just spent two months in three separate hospitals for family members. Diagnosis can be vastly increased in quality and sped up with proper databases and voice recognition.
How about you walk into a waiting room and for the hour you waste there get diagnosed by a intelligent computer expert system. Then when the doc gets around to you they do a separate diagnosis.
Those two will resolve to a set of probability-based diagnoses and automatic recommendations for further tests and questions. It can access a state-of-the-art managed database with references to relevant research and a place to enter commentary and questions.
If you fill out a questionaire, why wait for a human to do the Googling? Shouldn't the machine be better at it, and isn't Watson closer to this goal then ever?
No. Watson is a family of algorithms IBM uses. Not even a single machine, it's more of a brand name. And some of the impressive feats like Jeopardy are more like gimmicks than actual advanced AI.
Furthermore, actual medical training greatly favors rote memorization and application of standard operating procedures. As a computational biologist, I spend a fair amount of time thinking about probabilities and medically related topics. Over the years, I have had several frustrating conversations with doctors who refuse to consider priors or stats from well respected meta-analyses (e.g. by Cochrane reviews). But, who can blame them? Doctors have little time to think in the current system.
