A decision support system for diagnostic consultation in laboratory tests

1. INTRODUCTION. In primary care, the appropriate selection of test items is needed for correct diagnosis; it is also needed from the point of view of cost effectiveness. Thus, the Japan Society of Clinical Pathology recommended screening test items in clinical medicine [1]. In this study, we developed a microcomputer-based decision support system using these screening items to guide general medical practitioners to a correct interpretation of the results of laboratory tests. 2. SYSTEM Configuration The current version of the decision support system is designed to be run on PC-9821Bp (486 NEC computer) under the Window 3.0 environment. The input parameters are 26 essential items of laboratory tests, including urinalysis, fecal occult blood, hematological, and biochemical and serological tests recommended by the Japan Society of Clinical Pathology. The support system is composed of two processes. Firstly, the system can output any of the following diagnostic categories: 1) inflammatory disease, 2) muscular or myocardial disease, 3) anemia, 4) malignant tumor, 5) reno-urinary disease, 6) hepatobiliary disease, 7) diabetes mellitus, 8) gastrointestinal disease, 9) bone disease, 10) hyperlipidemia, and 11) normal. In the next step, the system can output some recommendations as to what minimum tests or what maneuver must be done to reconfirm the final diagnosis. The diagnostic part of the program uses decision-tree logic. 3. RESULTS. The subjects were 219 patients who were admitted to the Nagoya University Hospital; the diagnostic categories had been already confirmed. In each diagnostic category, 10-20 cases of infectious disease, 34 of 74 cases of malignant disease, 2 of 6 cases of muscular disease, 14 of 14 cases of anemia, 2 of 16 cases of malignant tumor, 42 of 55 cases of reno-urinary disease, 8 of 14 cases of diabetes mellitus, 9 of 20 cases of gastrointestinal disease, 3 of 26 cases of bone disease, and 14 of 18 cases of hyperlipidemia were correctly diagnosed using the first step of this system. 4. DISCUSSION. This diagnosis supporting system was very efficient for screening anemia, reno-urinary disease, diabetes mellitus, and hyperlipidemia. If this system is combined with information from disease history and physical examination, the diagnosis will be both easy and enhanced. Since the system can suggest effective tests to pinpoint the disease, unnecessary tests will be avoided. Lower diagnostic accuracy was seen in infectious disease, malignant tumor, muscular disease, and bone disease. For these diagnostic categories, additional items are necessary. In conclusion, basic tests are very useful for screening anemia, reno-urinary disease, diabetes mellitus, and hyperlipidemia; the system can provide the minimum and correct test items for diagnosis in primary care.