Lung Cancer Survival Time Prediction Using Machine Learning and Deep Learning Techniques

Accurate prediction of patient survival is pivotal for discerning prognostic indicators from radiological imagery. This study addresses the critical task of estimating the survival time of patients diagnosed with lung cancer. In this paper, we utilized Machine Learning (ML) and Deep Learning (DL) models to analyze three-dimensional radiological images. These models were deployed on datasets consisting of three-dimensional CT scan images, a set of pre-extracted quantitative imaging features, and clinical data. We evaluated nine distinct Machine Learning models, complemented by one Deep Learning model known as DeepSurv, and we used the Cox proportional hazard (Cox-PH) model—a semi-parametric survival model. Results indicate that, within the range of models evaluated, the Random Forest (RF) model exhibited the highest performance, achieving an accuracy of 0.699. The DeepSurv model followed with a score of 0.702, while the Cox-PH model outperformed all others with an accuracy of 0.719. This study underscores the efficacy of ML, DL, and Cox-PH models in forecasting the survival time of lung cancer patients using radiological images. These outcomes offer significant potential for the advancement of accurate prognostic models, which enhance patient care and treatment decision-making processes.