Statistical methodology for evaluating gastrointestinal symptoms
Spiegelhalter, D.J.
Clinics in Gastroenterology 14(3): 489-515
1985
ISSN/ISBN: 0300-5089 PMID: 4064353 Document Number: 3290
In this paper we have briefly indicated the power and intuitive appeal of using symptom information to make predictions concerning individual patients, from which it may be possible to allocate to risk groups in order to speed investigations for those that need them, and avoid unnecessary expenditure on those who do not. It may well be the case that the act of collecting defined symptom data in a structured manner is itself of considerable benefit, but it also seems appropriate to explore the extent to which reliable predictions may be made on the basis of symptoms alone. The important aspect of these formal aids to clinical judgment is that they are intended for use in a thoughtful and conscientious manner, and the aim in emphasizing scoring methods is that they make the process simple and transparent. The risk of implementation on computers is that the aids will be used in an unthinking manner or as a 'black box'. While it is probably good that a computer prediction should be viewed as 'just another test', it seems reasonable that clinicians should understand the basis for the computer judgement as much as they are supposed to understand the scientific background of a biochemical test.
Document emailed within 1 workday
