Small worlds and medical expertise: implications for medical cognition and knowledge engineering

This paper proposes and defends the small worlds hypothesis, which states that expert physicians organize diagnostic knowledge on the basis of similarities between disease categories, forming 'small worlds' consisting of small subsets of diseases and their distinguishing features. Examining existing data from several previous studies, the authors provide support for the small worlds hypothesis and for a characterization of the process of expert medical diagnostic reasoning as a succession of limited comparisons involving related diagnostic hypotheses. In one study, subjects were presented clinical endocrine cases one statement at a time and were prompted to think aloud after presentation of each statement. A combination of discourse and protocol analysis techniques were used to investigate hypothesis generation and evaluation. In another study, dialogues from doctor-patient interviews were examined. It was found that expert subjects rapidly select relatively small sets of plausible diagnostic hypotheses (small worlds) and focus on the most relevant medical findings that distinguish among the diseases in such small worlds. Results from both studies indicate that expert physicians use efficient strategies for discriminating among these alternative hypotheses in a stepwise process. In contrast, non-experts often generate large numbers of possible diagnostic hypotheses, belonging to widely differing disease categories. The results provide empirical support for the theoretical basis of small worlds. The implications of these results for the study of medical expertise and knowledge engineering are discussed, as well as considerations for the development of decision support systems.