A Proposed Model to Identify Factors Affecting Asthma using Data Mining

Introduction: The identification of asthma risk factors plays an important role in the prevention of the asthma as well as reducing the severity of symptoms. Nowadays, the identification process can be performed using modern techniques. Data mining is one of the techniques which has many applications in the fields of diagnosis, prediction, and treatment. This study aimed to identify the effective factors on asthma to provide a predictive model using data mining algorithms. Materials & Methods: This descriptive study with a practical approach included 220 data bases. The data were collected using a checklist and interviews from the patients referred to clinical centers of Shahid Sadoughi Hospital in Yazd, Iran, during 2014. The data were analyzed in SPSS IBM Modeler software (Version 14.2). Moreover, the CHAID decision tree,C5 algorithm, neural network algorithm, and Bayesian network algorithm were utilized in the modeling. Findings: In total, 12 variables were determined as the most influential factors in this study. The accuracy of the model on the data was estimated at 72.73%, 69.1%, 70.9%, and 65.45% in the CHAID algorithm, C5, Bayesian network, and the neural network, respectively. Discussion & Conclusions: According to the results, the performance accuracy of the model obtained from CHAID decision tree algorithm (73/72%) was higher than that of the other models. Moreover, an individual’s risk of asthma can be predicted with regard to the predictive factors and the established rules for a new sample with distinctive features.