Aggregating Abstaining and Delegating Classifiers For Improving Classification performance: An application to lung cancer survival prediction

The incidence of lung cancer is increasing, particularly among elderly patients with approximately 40% arising in patients over 70 years [1]. In 2002, approximately 6,700,000 death were due to cancer (3,796,000 men and 2,928,000 women), corresponding to the third cause of death [2]. Lung cancer has the highest overall mortality in the Western World with over 1 million deaths annually. Data extracted from Microarrays chips is considered to be an important source for providing insight about cancer. Particularly, when it concerns survival prediction outcome, several studies have reported on the successful application of supervised machine learning approaches to prediction of cancer [3-5].