Friday, February 25, 2011

MedInformaticsMD in WSJ again: No Patient Will Ever Say, 'Quick, Watson, the Needle!'

Following along the lines of my Jan. 2011 post IBM's Watson, Jeopardy, and "Revolutionizing Medicine" , the Wall Street Journal published this Letter to the Editor by me today:

Wall Street Journal

FEBRUARY 25, 2011

No Patient Will Ever Say, 'Quick, Watson, the Needle!'

Regarding Ray Kurzweil's "When Computers Beat Humans on Jeopardy" (op-ed, Feb. 17): Librarians often receive requests for information as cryptic as "Jeopardy!" clues, from people who are sometimes not even sure what they're seeking. Watson is, in essence, a librarian that retrieves facts.

Regarding natural language processing and fact-retrieval systems like IBM's Watson, medicine is about cognition. It's about human judgment born of experience in dealing with ambiguity, not just of language but also, and this is critical, of observations, findings, lab data, image interpretation, etc. It is about human intuition, assemblage and the integration of a huge amount of disparate information in ways not well understood even by its practitioners. The end result is not just the recall of a piece of information, obscure as the information might be.

I consider predictions of Watson spearheading cybernetic miracles in medicine to imply just as grandiose a valuation to the technology as the statements I heard two decades ago about the health information technology of the day, or even today, "revolutionizing medicine." That has not happened.

A cybernetic librarian is no physician. As for me, for now I'll stick to people in my own medical care. Let Mr. Kurzweil see the machine for his ailments.

Scot Silverstein, M.D.


I am pleased that my counter-marketing-hype opinion on the meme of health IT "revolutionizing medicine" was deemed fit to print by the Journal.

Those in senior leadership positions, as are included in the WSJ's readership, need to see that meme challenged.

-- SS


Anonymous said...

This letter contains a fabulous description of the work of the effective clinician, and why the notion that Watson can substitute is folly. There are doctors who I have met who can recite facts like there is no tomorrow, but are useless as clinicians because of their failure to be gifted in the characteristics described in the letter. Thank you.

InformaticsMD said...

I contend that any clinician with a Medical Informatics or related background who's developed an information system meant for complex clinical settings should understand this description.

The mono-disciplinary computer scientists and computer technicians making predictions of cybernetic miracles in medicine, sometime soon, don't know what they don't know.

They're good at writing arcane papers filled with the latest, greatest algorithmic advances. Those "advances" translate into some incremental performance increase on some obscure metric, although usually under "special circumstance", and needing "further research" for wider applicability.

-- SS

Anonymous said...

Quick, Watson, find a nurse; and find a doctor with judgment. I am suffering from erroneous esoteric diagnoses and attendant tretament errors.

InformaticsMD said...

I wonder why there are no suggestions of replacing lawyers with computers.

Perhaps Watson could be programmed to, say, determine if a proposed law is constitutional or not?

Like medicine, that should be child's play, no?

-- SS