Prediction models applied to lung cancer using data mining

Abstract

Lung cancer is the most common cause of cancer death in men and the second leading cause of cancer death in women worldwide. Even though early detection of cancer can aid in the complete cure of the disease, the demand for techniques to detect the occurrence of cancer nodules at an early stage is increasing. Its cure rate and prediction are primarily dependent on early disease detection and diagnosis. Knowledge discovery and data mining have numerous applications in the business and scientific domains that provide useful information in healthcare systems. Therefore, the present work aimed to compare several prediction models as well as the features to be used, with the help of Weka and RapidMiner tools. Both classification and association rules techniques were implemented. The results obtained were quite satisfactory, with emphasis on the Naive Bayes model, which obtained an accuracy of 95.03% for cross-validation 10 folds and 94.59% for percentage split 66%.