Home

Fast Classification of Pathological Processes in Peritoneal Dialysis Patients Based on Blood Samples


  Daria Prilutsky [1,2,3]  ,  Boris Rogachev [4]  ,  Marina Vorobiov [4]  ,  Leslie Lobel [1]  ,  Mark Last [3]  ,  Robert S. Marks [2,5,6]  
 [1] Department of Virology, Faculty of Health Science, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel
[2] National Institute of Biotechnology in the Negev, Ben-Gurion University of the Negev, P. O. Box 653, Beer-Sheva 84105, Israel
[3] Department of Information Systems Engineering, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel
[4] Department of Nephrology, Soroka Medical Center, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel
[5] Department of Biotechnology Engineering, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva, 84105, Israel
[6] The Ilse Katz Center for Meso and Nanoscale Science and Technology, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel

Peritoneal dialysis (PD) is a recognized form of treatment for patients with severe chronic kidney failure. Acute peritonitis is a major complication associated with this procedure, while frequent occurrence of peritonitis can be associated with high risk of mortality and morbidity. Diagnosis of peritonitis is heavily dependent on laboratory tests including etiological agents’ detection, but these time-consuming tests are subject to accuracy and sensitivity limitations.  In our research, a fast diagnosis of bacterial peritonitis is based on polymorphonuclear leukocytes (PMNs) functional changes that can be assessed by the chemiluminescent (CL) reaction within several hours.  For the assessment and discrimination of functional states of PMNs, we have applied a classification algorithm (C4.5) to the time-series data produced by dynamic component chemiluminescence analysis in a luminol-amplified whole blood samples taken from PD patients. Feature extraction is based on the evaluation of CL kinetic patterns of stimulated PMNs. This novel method of combining whole-blood CL, kinetics component analysis, and classification of clinical groups using a decision-tree algorithm has shown a high predictive diagnostic value and it may assist clinicians in detection of PD associated clinical states.