Evaluation Of The Phadiatop Test Results Using Chaid Algorithm And Logistic Regression

This paper presents the use of logistic regression and decision tree (CHAID algorithm) for classification of patients into one of the two groups of Phadiatop test. Our data come from patients who underwent Phadiatop test examinations in the University Hospital of Ostrava, the Department of Work and preventive medicine. As predictive variables for discriminant analysis, personal and family anamnesis, sex and age of the patient were selected and appropriately rated. These data were obtained on the spot assessment of the manifestations of atopy. Both of these anamneses were divided into four categories according to severity ranked by doctors (asthma, allergic rhinitis, eczema, and others). Evaluation was based on the test results. One of two possible values is involved there: 0 for Phadiatop test result 0 and I, or 1 for values 11 to VI. Both methods provide very similar results with the success rate 64% for correct classification of patients. Model was tested using a database of 274 clients. Both methods used statistically significant diseases which influence the Phadiatop test results - asthma and allergic rhinitis in personal anamnesis.