IBaI Institute |
Home |
Past events |
ICDM´2001 |
Abstracts
1st Industrial Conference on Data Mining ICDM´2001
24. 07. - 25. 07. 2001 Leipzig/Germany
Evaluation of Clinical Relevance of Clinical Laboratory Investigations by Data Mining
Dr. med. Ulrich Sack, University of Leipzig/Germany
The diagnostic investigation of immunologically influenced diseases includes the determination of serological and cellular parameters in the peripheral blood of patients. For the detection of these parameters, a variety of well established and new fashioned immunoassays are available. Since these test kits have been shown to yield highly different results of unknown clinical significance, we have compared a selection of commercial test kits and have analysed their diagnostic value by data mining. Here we describe applications of data mining for the diagnosis of inflammatory and thrombotic induced acute central nervous processes and identification of various prognostic groups of cancer patients. Evaluation of laboratory results by data mining revealed a restricted suitability of chosen test parameters to reply diagnostic questions. Thereby, unnecessarily performed test systems could be removed from the diagnostic panel. Furthermore, computer assisted classification in positive and negative results according to clinical findings could be implemented.