Training the Expert Mind, Part II: Medical Diagnosis

In Part I, I gave a brief review of an article in Scientific
American, entitled href="http://www.sciam.com/print_version.cfm?articleID=00010347-101C-14C1-8F9E83414B7F4945">The
Expert Mind.
 The article described the outcome of research into the mental
processes of expert chess players.  The motivation for the
research is to find out how expertise works, to see if there is a
systematic way to develop expertise in a variety of fields.



Perhaps the most important finding in this endeavor is that experts
have developed, through a great deal of practice, an excellent capacity
for pattern recognition.  The recognition of patterns is, of
course, what medical diagnosis is all about.

Expert chess players use their ability to recognize abstract patterns
to narrow down the possible moves to consider.  



Likewise, expert physicians go through a similar process, when
establishing a diagnosis.  Like a chess player who can see
that there are certain sequences of moves that are not worth pursuing,
an experienced doctor can rule out entire categories of illness, and
not consider them any further (for a while, anyway).  



I was planning to tie this to an article that I read a while back, in
which the authors studied the thought process of experienced clinicians
in a grand-rounds-type of situation, and compared that to the process
that medical students and residents used in the same cases.  I
think tt was in href="http://www.academicmedicine.org/pt/re/acmed/home.htm">Academic
Medicine, back in about 1991 or 1992.  



Unfortunately, the only keywords I can think of to search on, are words
like "diagnosis."  Searching a medical journal for the word
"diagnosis is rather pointless, obviously.  I paged through
the tables of
contents of 1991.  Their database started giving "null"
responses for the issues in 1992, and since I am not entirely sure I
was looking at the correct journal, I gave up.  



I admit, this post would be a lot more interesting if I could find that
article.  If anyone else can recall it, please let me know.



Anyway, the main point was that the more experienced clinicians would
use incisive questions to narrow down the list of possible diagnoses in
broad strokes.  Because of their experience, they could choose
questions according to an efficient search algorithm.  Less
experienced clinicians tended to fumble around a bit more, going down
unproductive paths.



Even though it was not the article I was looking for, I did find one
that was relevant:

href="http://www.academicmedicine.org/pt/re/acmed/abstract.00001888-199110000-00015.htm">The
characteristics of early diagnostic hypotheses generated by
physicians (experts) and students (novices) at one medical school.




Academic Medicine. 66(10):607-12, October 1991.

Sisson, J C; Donnelly, M B; Hess, G E; Woolliscroft, J O

Abstract:

How hypotheses generated at the outset of a physician-patient encounter
influence the physician's diagnostic reasoning has received relatively
little attention in the literature. Yet, this activity modulates the
efficiency and accuracy of the diagnostic process. The authors proposed
the theory that physicians (experts) would generate less specific
initial diagnostic hypotheses than would students (novices). In 1988
and 1989, the hypotheses generated by 32 practicing physicians and 39
third-year medical students at one medical school were analyzed for
number, specificity, and breadth (i.e., range) of diagnoses, and for
differences between those generated by the physicians and by the
students. The hypotheses were formulated on the basis of the initially
available data-chief complaint and its duration, and basic demographic
descriptions-from patients in three clinical scenarios. Compared with
the physicians, the students generated significantly more hypotheses
and significantly more specific hypotheses. There was no significant
difference between the groups in the breadths of the hypotheses they
selected, that is, the numbers of broad categories of diagnosis covered
by individuals' hypotheses.

(C) 1991 Association of American Medical Colleges 

I'm somewhat sure that this is not the article I was looking for,
because it was written by my former professors.  I probably
would have
remembered that. 



Sission et. al. found that less experienced
clinicians actually generated more hypotheses.  At first
glance, that would seem to indicate that the third-year medical
students are better diagnosticians.  Perhaps that is true, but
the article does not say.  I would prefer to believe that it
is not the case; rather, the more experienced clinicians generated
fewer hypotheses because they were using a more efficient search
algorithm.  


The findings of the research on chess players is consistent with
that.  



What these articles do not tell us, is whether there is a quick way to
teach medical students to be more efficient.  The findings
regarding chess players would seem to indicate that there probably is
not.  It still takes years of study of individual cases in
order to develop this skill.