Computer Programs to Support Clinical Decision Making

Abstract

To the Editor.— In his excellent review of the current progress in computerized medical expert systems, Dr shortliffe 1 neglects an important human factor issue: familiarity. No matter how sophisticated, tools can only become useful when we learn their strengths and weaknesses and discover how they work in a number of different situations. For this reason, a program such as DXplain 2 may never find wide acceptance. Because it is intended to help the physician only with particularly difficult diagnoses, it is unlikely that he or she will ever gain enough experience to be able to use it with confidence. Although desirable, the ability of the program to review its reasoning may not necessarily substitute for experience gained with the program under a variety of circumstances. It is unfortunate that, in his editorial in the same issue of JAMA , Mr Harris 3 chose to reinforce the fear that physicians may