A knowledge-based system to aid with the clinical interpretation of complex serum protein data

In every area of science workers are finding increasing difficulty in managing the volume of available data. In medicine, the accelerating pace has the worrisome overtones of our failing to provide up-to-date care for our patients. In other information-intensive areas, we rely heavily on software that manages much of the complexity unseen. Patient care could benefit enormously from the incorporation of "knowledge-based" programs to aid with diagnosis and management of many disorders. This article describes such a system designed to organize complex data which can be viewed as a test cluster aimed at many disorders pertinent to serum proteins. This program performs complex tasks such as reference range adjustment, ICD-9 code assignment, and searching for diagnostic "signatures", to generate clinically relevant text and simple graphics. The results have been remarkably accurate and produce repeatable results at the rate of approximately 10 cases per minute. The reluctance to embrace software assistance in laboratory medicine may have serious consequences in the short term and disastrous results within a decade. Expanding the limited algorithm described here to include more traditional chemistry testing could provide the very assistance that all in clinical care desire, a laboratory tool as powerful and adaptable as the traditional physical exam.