Predicting lung cancer survivability: A machine learning regression model

Abstract

Lung cancer is one of the main leading causes of cancer death in all over the world. Accurate prediction of lung cancer survivability can enable physicians to make more reliable decisions about a patient's treatment. The objective of this research is to design robust machine learning model with supervised regression model to predict survivability of the lung cancer patients. This work includes Multiple Linear Regression, Support Vector Regression with Radial Function, Random Forest, Extreme Gradient Boosting Tree regression algorithms to build an ensemble model using stacking technology with meta-learner Gradient Boosting Machine. This experiment is performed on large SEER 2011-2017 dataset. The novel model achieved a high root mean squared error (RMSE) value of 8.58459 on the test dataset which outperforms the base models. The experimentation results show that the proposed system attains better result compared to the existing models.

Keywords

• lung cancer
• regression
• stacking technology
• ensemble
• extreme gradient boosting tree