Application of Knowledge Discovery in Database to Blood Cell Counter Data to Improve Quality Control in Clinical Pathology

Abstract

This paper analyzes the functions of an automated blood cell counter from a clinical pathology laboratory and proposes an application of the data preprocessing phase of the Knowledge Discovery in Databases (KDD) processes such as data cleaning, data selection and data transformation on the blood cell counter data to enhance the quality of the system. The blood cell counter data are subjected to a range test normally to ensure quality and then the tests are repeated manually to ensure correctness of the results. The manual procedure is time consuming and hence it is desired to reduce the number of manual reviews. A delta check is performed in some cases to reduce the number of manual reviews and that involves the previous test results of the same patient for the same test. This paper analyzes various ways of finding the delta values of the tests to reduce the number of manual reviews. Two exceptions to the process are also analyzed in this paper. The first case is to repeat the test even when the test result is normal and the second is not to go for retest even if the test results are abnormal based on the past results of the same patient.