Time-Based Survival Analysis for Breast Cancer

The aim of this study is to predict the survival rates of breast cancer patients using the Cox regression and Kaplan–Meier model implementation and perform a comparative analysis against other ML models (Li et al. in PLoS ONE 16, 2021 1; Lou et al. in Cancers (Basel). 2020;12(12):3817 1). This study attempts to study the survival rates of breast cancer patients given by the University of Chicago’s Billings Hospital dataset. This study is done on a time analysis survival method as opposed to existing medical research to detect cancer in patients. The machine learning method used to achieve the same is Cox regression and Kaplan–Meier estimator. Time parameters such as the age of the patient, time of operation/treatment, and biological aspects such as the number of positive axillary nodes detected have been taken into account before performing the Kaplan–Meier estimator and the Cox proportional hazards model that provides the survival curve function and provides a probabilistic value for whether the patient will survive more than a given threshold time. Other conventional methods have also been discussed in order to analyze and prove that Cox regression is the most suitable method for the given parameters and result required. Finally, it is observed that the CoxBoost applied on the Cox regression model provided the most accurate outcomes.