ABSTRACT

Diagnosis is one of the doctor's central tasks. As an intellectual activity, it is a variant of the more general skill of classification—assigning entities to different classes or categories. Classification is easy when each category (or, in the present case, each disease) has a specific, reliably detected, sign. Unfortunately, such 'pathognomonic' signs are rare. The common signs of illness (such as fever or pain) are shared by many different diseases and most laboratory test results have more than one possible cause. This non-specific relationship between signs and diseases ensures that there will always be an element of uncertainty in medical diagnosis. This uncertainty can be reduced by the discovery of more sensitive clinical signs, but it can never be eliminated. No test is perfectly accurate, signs can be misleading, and even the best treatments do not always succeed (see Schwartz and Griffin, 1986, for more on the probabilistic nature of medical decision-making).