A data mining algorithm for determination of influential factors on the hospitalization of patients subject to chronic obstructive pulmonary disease

Background: The present study is on the development of a data mining algorithm for finding the influential factors on the hospitalization of patients subject to chronic obstructive pulmonary disease. Materials and Methods: This is a descriptive analytical study conducted cross sectionally in 2017 on a research community of 150 people with disease symptoms referred to clinics and hospitals across Tehran (Iran). The people were surveyed by a self-designed questionnaire, including queries on life style and family information. The sampling was simple intuitive from previously published studies. The modeling of the data was based on the CRISP method. The C5 decision tree algorithm was used and the data was analyzed by RapidMiner software. Results: The common symptoms of the patients were found to be shortness of breath, cough, chest pain, sputum, continuous cold, and cyanogens. Besides, the family history, smoking, and exposure to allergic agents were other influential factors on the disease. After accomplishment of this study, the results were consulted with the experts of the field. Conclution: It is concluded that data mining can be applied for excavation of knowledge from the gathered data and for determination of the effective factors on patient conditions. Accordingly, this model can successfully predict the disease status of any patient from its symptoms.