Intelligent approaches for prognosticating post-operative life expectancy in the lung cancer patients

Abstract

The aim of this research is to evaluate the performance of two feature selection methods on seven different machine learning methods applied over thoracic surgery data. Feature selection is a crucial pre-processing step in determining factors responsible for post-operative life expectancy in the patients suffering with lung cancer. Postoperative life expectancy complications are the most common fatality following major types of thoracic surgery. In particular, we want to examine the underlying health factors of patients that could potentially be a powerful predictor for deaths which are surgically related. Seven machine learning methods namely Naïve Bayes, Linear SVM, MLP, RBF Network, SMO, KNN and CART are employed for analyzing the performance of feature selection methods. Maximum accuracy of 85.11% was obtained with correlation-based feature selection in comparison with consistency-based feature selection which was 84.89 %.