Investigation of Factors Affecting Immunotherapy Treatment Results by Binary Logistic Regression and Classification Analysis

Abstract

There are many factors that affect the success of immunotherapy treatment. In addition to the clinical examination, the investigation of these factors by different methods contributes to the researchers on prior knowledge and time. In this study, it was aimed to evaluate the application of logistic regression and data mining methods to evaluate the success of post-immunotherapy treatment. Bilateral logistic regression analysis with WTA and classification analysis with Weka were used to evaluate whether immunotherapy treatment was successful for warts. Decision tree structure is also discussed to determine the variables that affect classification success. According to the logistic regression result, the model is important because the probe. 0.022 <0.05. The classification result for the logistic regression model was calculated as 85.56%. This result shows that the model is successful. Data mining experiments were carried out with different classification algorithms. The best result was found in decision trees (with J48 algorithm) with 85.56% accuracy rate. According to the J48 algorithm decision tree structure, the variables that affect the outcome of the treatment were recorded as time, number of warts and age, respectively. Study results show that both methods yield parallel results. Decision tree algorithm is used as an alternative to classical statistical models. In particular, in cases where clinical research is limited, it will benefit researchers on topics such as transition to analysis, preliminary information gathering, time and effort.