Prediction of Pneumoconiosis Disease Using Machine Learning Techniques

Abstract Pneumoconiosis (Silicosis) is considered to be lung disease and it is a common one in Egypt as its predominance rate ranges from 18.5 % to 45.8% among workers who are at risk of free crystalline silica dust exposure. The main reason behind its occurrence is the inhalation of dust, often in mines. The aim of this paper is to predict the Pneumoconiosis disease among the workers by the use of machine learning techniques as Pneumoconiosis can be prevented but not considered to be a treatable disease. We use principle component analysis algorithm to remove the redundant attributes, and then we use k-nearest neighbor and neural network classifiers in the classification phase. Results show that the weighted k-nearest neighbor classifier outperforms a wide variety of the neural network classifier.