[Dynamic generation of diagnostic knowledge from a database of well defined case clinical records]

To promote an evidence-based practice of laboratory medicine, we created a database of 1584 typical case records consisting of 30 common diseases. Cases without therapy or major complication were carefully selected. Characteristic clinical symptoms and signs indicating the severity were recorded together with general laboratory test results. A user-friendly computer system for interpretive laboratory diagnosis was also created. It features (1) automatic graphics for visualizing distributions of laboratory values and/or prevalence of clinical symptoms and signs, (2) simultaneous display of distributions of common test values for any combination of diseases, (3) quick subclassification of cases within a disease group by any clinical characteristic, (4) diagnostic guidance by dynamic calculation of a novel similarity index that represents matching of a given set of signs/values with those of typical cases in the database.