Analyzing the impact of epidermoid and adeno tissue on cancer incidence whit a data mining approach

Abstract

Early detection is the only way to effectively control diseases whose treatment can be challenging, expensive, and time-consuming. Identifying the influencing factors in the occurrence of disease can, therefore, reduce the time associated with diagnosis and provide a solid foundation for improving the prognosis and preventing patients' deterioration. Applying data mining techniques as a novel approach to the early detection of disease-causing agents can significantly assist the early detection. In this study, an attempt was made to investigate the effect of epidermoid and adeno tissues on the incidence of cancerous diseases such as bone, bone marrow, lung, and neck cancer by conducting a data mining process on cancer patient data sets. Hence, Implementing two data mining techniques, K-nearest neighbor and decision tree, on the data of patients with these four types of cancer, an attempt was made to evaluate their performance using the three criteria of accuracy, error ratio, and negative prediction value. The implementation of data mining techniques and evaluations of their performance indicates that the decision tree technique performed better with an accuracy of 89.10%, an error ratio of 14.04%, and negative prediction value of 77.71%. Also, based on the findings, contamination of epidermoid and adeno tissues does not affect the early detection of any of the four categories of bone, bone marrow, lung, or neck cancer. In other words, the infection of the two epidermoid and adeno tissues cannot be the cause of the four types of bone, bone marrow, lung, and neck cancer.