Logistická regrese jako nástroj pro diskriminaci v lékařských aplikacích

Abstract

This thesis is investigativly engaged in analysis of medical data derivable from surgical intervention colectomy in years 2001 – 2006. These data was first transformed into form of surgical factors and physiological factors and finally these data was used for generation of logistic regression model purposes that was verified by standard statistical by way of supplementary reference verifications technique. As a supplementary verifications technique was used an exponential analyses in improvement, which were algorithmized. Obtained logistic regressive models were confronted by supplied referential model. Found logistics regressive models are finally utilized for discriminating analysis purposes that consist in possibility of unknown pertinence of patient which is formate into one of two groups according to surgical complications. In the sample of delivered data is performed prognosis of surgical complications with errors evaluation of formatting correctness. The process found for this patient formatting including mistake evaluation is algorithmic processed in MATLAB.