Bayesian classifier based on cancer prognostic markers using accelerated failure time model with frailty effect

This work presents a Bayesian classifier technique to categorize patients based on predictive biomarkers of time-to-event data utilizing the Accelerated Failure Time (AFT) model incorporating the frailty effect. Before classification, efficient and significant markers from a high-dimensional gene expression dataset need to be identified. Currently, it is an emerging area in oncology. A conventional three-step feature selection approach is introduced to select the most relevant markers from high-dimensional data. A threshold value for each selected marker is obtained using the proposed Bayesian classification procedure incorporating the AFT model with the frailty effect. The frailty effect is incorporated to account for unobserved heterogeneity in the expression values of subjects, allowing for a more accurate investigation of the risk of cancer-related outcomes. A simulation study is performed to validate the proposed classification approach, and the classification’s performance is evaluated using the Brier score, and the result indicates that it is relatively high. Application of the proposed classification technique is provided for two high-dimensional TCGA datasets of lung cancer patients. Our results provide evidence about the effect of classified gene expression values on the survival risk of lung cancer patients.