A New Approach for Lung Cancer Classification using GSA and FCM with an Enhanced Fitness Function

Currently, the lung cancer is considered as one of the most leading cause of death in both men and women in worldwide due to health related issues. The current medical industry has developed several imaging technique to diagnose the lung cancer. However, the survival after diagnosis is a challenging problem. Thus, prediction of cancer in its early stage is a promising solution to prevent the mortality due to cancer. Smoking is the leading cause of cancer. There are several parameters which can be analyzed to identify the cancer in early stage such as smoking. In order to analyze these parameters, data mining is considered as the best solution. This work presents a data mining based machine learning approach for lung cancer prediction. The conventional data mining applications suffer from the issue of handling the missing values and selecting the significant features. Hence, missing value imputation and a new approach for feature selection using gravitational search optimization is also introduced. Moreover, this optimization process uses a transfer function and mutual information of features to design the new fitness function. Finally, Fuzzy C-means based unsupervised learning scheme is applied to learn the selected attributes. When compared to existing machine learning algorithms, the comparative research shows that the proposed technique outperforms them in terms of classification accuracy, precision, and recall.