The Division of Computer Medicine at Boston's Beth Israel Hospital has developed a clinical computing system which has accumulated data on more than 500,000 patients. I propose to develop medical decision support computer programs which help select test and management strategies in a variety of medical conditions. When possible, these programs will be based on data from our large patient database. The techniques of decision analysis (1,2) will be used to help structure and evaluate clinical problems. By having a large patient database linked to decision analysis programs, local population bias will be accounted for, and local clinical expertise will be factored into the decision analysis, rather than diluted out of the analysis. It is my hypothesis that the evailability of flexible decision analysis programs on our hospital systems will spur the development of a library of decision trees of common clinical problems, developed and approved by expert clinicians. These analyses could then be the basis of decision support aids on our hospital-wide system. I believe that the use of decision-supported test strategy selection will reduce the number of diagnostic tests ordered for many medical problems and consequently reduce patient morbidity, mortality, and hospitalization costs. These programs will have the additional benefit of teaching house staff how to order and interpret diagnostic tests.
Showing the most recent 10 out of 12 publications