Standardizing clinical laboratory data for the development of transferable computer-based diagnostic programs

Abstract

The existence of systematic differences between test results obtained at different laboratories can compromise the development of generally accessible reference databases for interpretive pathology. We review approaches to the elimination of inter-laboratory bias from pathology test results through the use of standard unit transformations. A general transform procedure is described that will permit laboratories serving a common population to make use of reference data, decision rules, and computer-based interpretive programs developed around a larger clinical database than each of these test centers could amass for themselves.