Predicting Lung Cancer Survival Prognosis based on the Conditional Survival Bayesian Network

Abstract Lung cancer is one of the leading causes of cancer death and impose an enormous economic burden on patients. It is important to develop an accurate risk assessment model to determine the appropriate treatment for patients after the initial diagnosis of lung cancer. The Cox proportional hazards model is mostly utilized in survival analysis. However, real-world medical data is always incomplete, which poses a great challenge to the application of the Cox proportional hazards model. The commonly used imputation methods cannot achieve sufficient accuracy in the issue of missing data, which drives us to investigate the novel imputation methods for the development of clinical prediction models. In this article, we present a novel missing data imputation method: Bayesian networks for inferring missing covariates. We collected a total of 5,240 patients diagnosed with lung cancer from Weihai Municipal Hospital, China. Then we applied a joint model that combined a Bayesian network and a Cox model to predict mortality risk in individual patients with lung cancer. The established prognostic model achieved a good predictive performance in discrimination and calibration. Through experiments, we proved that the Bayesian network methodology is a robust and accurate approach to addressing the issue of missing data. We showed that combining the Bayesian network with the Cox proportional hazards model is highly beneficial, providing a more efficient tool for risk prediction.