Indian Journal of Science and Technology

Abstract

Abstract Background: Decision trees are a simple and powerful form of multiple variable analyses which allows predicting, explaining, describing, or classifying an outcome. The risk factors for Adverse event include demographic features of patients and concurrent illnesses, hypersensitivity to related drugs, drugs currently taken etc. Methods: The objective is to classify the Adverse event Thyroid cancer outcomes based on the risk factors. For that Decision tree based classifier model that uses Naïve entropy for calculating the information gain is proposed for classifying the adverse event .First the missing values in the dataset are handled using mean of nearby points. Along with that, Association function is used for determining the relative degree between the given attribute and class C. Findings: The proposed Classifier Model generates the If then Rules for adverse event outcome of Thyroid cancer. It considers only three attributes such as age, drug name and indication for using the drug for generating the tree structure. Hence the depth of the tree is reduced. The rules generated are grouped into categories .Then they are arranged in the descending order based on the number of occurrences of the rules. The top rules specify the major occurrence of adverse event for the combination of the attribute valus.Accuracy of the proposed classifier model is compared with that of J48, KNN and Naïve Bayes algorithm. The proposed model has better accuracy than that of other classifiers. Application: The different rules generated are stored in the database. In order to prevent adverse event thyroid cancer, the physicians can make use of this database and avoid medications in these combinations.

Keywords: Adverse Event, Association Function, Data Mining, Decision Tree, Entropy