Abstract:Effective management of colorectal cancer (CRC) necessitates precise prognostication and informed decision-making, yet existing literature often lacks emphasis on parsimonious variable selection and conveying complex interdependencies among factors to medical practitioners. To address this gap, we propose a decision support system integrating Elastic Net (EN) and Simulated Annealing (SA) algorithms for variable selection, followed by Tree Augmented Naive Bayes (TAN) modeling to elucidate conditional relationships. Through k-fold cross-validation, we identify optimal TAN models with varying variable sets and explore interdependency structures. Our approach acknowledges the challenge of conveying intricate relationships among numerous variables to medical practitioners and aims to enhance patient-physician communication. The stage of cancer emerges as a robust predictor, with its significance amplified by the number of metastatic lymph nodes. Moreover, the impact of metastatic lymph nodes on survival prediction varies with the age of diagnosis, with diminished relevance observed in older patients. Age itself emerges as a crucial determinant of survival, yet its effect is modulated by marital status. Leveraging these insights, we develop a web-based tool to facilitate physician–patient communication, mitigate clinical inertia, and enhance decision-making in CRC treatment. This research contributes to a parsimonious model with superior predictive capabilities while uncovering hidden conditional relationships, fostering more meaningful discussions between physicians and patients without compromising patient satisfaction with healthcare provision.

A Parsimonious Tree Augmented Naive Bayes Model for Exploring Colorectal Cancer Survival Factors and Their Conditional Interrelations