DESIGN CHAIN     Today's Designs for Tomorrow's Innovations

I'll Take Watson for 2014, Alex

NO RATINGS
View comments: newest first | oldest first | threaded
Rich Krajewski
User Rank
Supply Network Guru
Re: amazing
Rich Krajewski   2/5/2014 2:13:28 AM
NO RATINGS

"I doubt it will ever replace a human completely as the ultimate opinion."

But why do you have that opinion? Computers run some sophisticated financial trading operations. If, as forms of AI, they manage to zero in on high-probability diagnoses in medicine, they may quickly find a pattern that creates a high degree of success.

My worry is that this pattern might be devastating for those on the "tail ends" of the bell curve (that is, those who have more esoteric problems), in essence locking them out from effective medical treatment, as computer-generated diagnoses become best practices, and anything else is considered to be experimental, or reckless. That pattern may prevent physicians from attempting diagnoses that fall too far afield from the computer norm.




FLYINGSCOT
User Rank
Supply Network Guru
amazing
FLYINGSCOT   2/4/2014 10:38:03 AM
NO RATINGS

I remember watching that clip of Watson on Jeopardy.  It was extremely impressive.  If we can harness this technology to assist the medical community we should see a great improvement in overall patient care but I doubt it will ever replace a human completely as the ultimate opinion.

Rich Krajewski
User Rank
Supply Network Guru
Not to
Rich Krajewski   2/3/2014 9:05:56 PM

Not to disparage Watson, but I wonder if it, as a representative of bell-curve medicine, is already on its way out. What I mean by bell-curve medicine is the application of highest probability statistics, which depends on groups, to diagnoses of individuals. The day of personalized medicine is fast approaching, where it will no longer matter that 95% of the population responds favorably to a particular treatment, when my genetic makeup shows that I am among the 5% who do not respond, or, worse, respond with seriously adverse effects.

Moreover, is Watson likely to see applications of medical research that seem completely outside the area of the original research? For instance, would Watson have been able to synthesize the possibility that anti-depressant medication at doses that are sub-therapeutic for psychiatric uses would be useful as a treatment for symptoms of bowel inflammation? I would be surprised if that were so.

On the other hand, if Watson were programmed to take genetic polymorphisms into account, or epigenetic differences in protein expression, then it would no doubt be a fantastic tool for physicians. But, only a tool. I would hate to see insurance companies, hospitals, and medical associations insisting that physicians follow the bell-curve suggestions of a program like Watson, for liability or pseudo "best practices" reasons.



Latest Poll
EBN Newswire
Twitter Feed
EBN Online Twitter Feed