A combined iterative sure independence screening and Cox proportional hazard model for extracting and analyzing prognostic biomarkers of adenocarcinoma lung cancer

Abstract

The selection of significant biomarkers is essential in researching cancer diagnosis and treatment. The independence screening method works substantively to select crucial features based on the conditional marginal selection method. But it may pretermit the concoction effect of some marginally less essential covariates. We aim to obtain a significant biomarker-specific prediction on overall survival to know their survival and death risk. In this work, an iterative sure independence screening (ISIS) scheme has been applied to extract features from the high-dimensional dataset of adenocarcinoma lung cancer. Conventional and Bayesian approaches of the Cox proportional hazard (CPH) model have been used for analyzing the data to provide interpretation and conclusions about survival estimates. The accelerated failure time model is also used as an alternative to the CPH model. A forest plot is employed to show the graphical representation of the meta-analysis of the study design. Utilizing ISIS, we selected up to 20 relevant features From the entire dataset of adenocarcinoma lung cancer; some of them are liable to produce a positive hazard ratio greater than 1, and some are less than 1. The P values associated with the selected biomarkers imply their statistical significance. Fourteen biomarkers have been identified with a hazard ratio of less than 1; the remaining 20 biomarkers are greater than 1. These 14 biomarkers produce less risk of death for patients with adenocarcinoma lung cancer, and the remaining six biomarkers result in a high risk of death from adenocarcinoma lung cancer.