Thursday, July 07, 2005

New idea - online robot doctors

[See previous]

The idea of an "expert system" for medical diagnosis has been around for a long time. But it has been hard to find public examples. Here's a simple one that has come online recently in the "vitamins and minerals" space.

Health Web site launches online diagnosis

From the article: This is an incredibly primitive system (not an expert system at all...) but it is getting press and it definitely shows that people are interested.

What this has me wondering is the following -- could several teams or companies (Google? Microsoft? WebMD?) put full-blown digital doctors online to analyze symptoms and attempt diagnoses? These systems could compete against each other in the Grand Challenge sense and improve over time, until we come to the point that they are right 50% of the time, then 60% of the time, then 70% of the time...

Eventually these systems would become better than human doctors, because they would be able to account for all sorts of drug interactions, side effects, etc. that the human brain cannot manage. These robotic doctors would eventually replace human doctors and improve health care in the process.

We've been hearing about this as a possibility for 30 years. How long will it be before something like this actually happens?

Comments:
This isn't an expert system any more than your page is an expert system on making money. This is just a web page with (possibly) useful information.

MYCIN was an early medical diagnosis expert system, and you will learn about it in any undergraduate AI class. Research continues in expert systems. The biggest cost is gathering the expert knowledge. That cost could be reduced by creating system that can read texts or learn from experiential evidence. AI researchers have been working on those problems for decades too. There may be a huge breakthrough just around the corner, or we could keep getting incremental improvements for several more decades.

It is more likely that human doctors in cheaper economies would be used as cheap expert systems in the near future.
 
Marshall is right, you could save hundreds of dollars with this system instead of going to the doctor. Obviously people won't be able to diagnose every problem, but if I had a chance to save a few hundred bucks I would do it in a heartbeat.
 
I think the reason this has not been done so far has to do with liability issues. None of the big companies probably want to risk the lawsuits that would inevitably pop up. Of course, maybe some enterprising insurance carrier could start offering "internet malpractice" coverage?

Another issue seems to be that doctors as a whole seem to be fairly net illiterate (at least the ones I know). Other than playing golf video games, my sense is that the majority of docs leave the "techie" stuff to their staff or kids. I'm still looking for a doc that I could email a quick question to or schedule my appointment on-line.

I think that a paradigm shift could happen if some young, tech savvy doctor wrote a "best practices" book that showed doctors how to use computer and internet technology to their advantage, and how they might be able to actually make money and serve patients better.

In answer to Marshall’s question: 5 years.
 
Hi there. I like your blog. I was just blogging and thought i'd leave a comment on it. Have a good one.

regards,
health insurance quote for child
 
Post a Comment

<< Home
ARCHIVES © Copyright 2003-2005 by Marshall Brain

RSS

This page is powered by Blogger. Isn't yours?