Risk Factor Prediction and Survival Analysis for Chronic Disease Recurrence Using Machine Learning Models

Abstract

Chronic diseases are a major public health concern in KSA, reflecting global trends in prevalence and impact. Among these, ccolorectal cancer (CRC) is one that is most urgent, being the most frequently diagnosed cancer in males and the third most common in females in KSA. Our research introduces a novel approach to addressing the pressing issue of chronic disease recurrence, specifically in CRC, by applying advanced machine learning models for risk factor prediction and survival analysis. Through this innovative methodology, we anticipate significant contributions to CRC management and prevention, with the potential to transform patient care by offering more accurate and individualized predictions. Expected outcomes include the improved identification of high-risk individuals and a deeper understanding of recurrence patterns, which could guide tailored interventions and enhance overall healthcare strategies for CRCThe study will use descriptive statistics to summarize participants' clinical and demographic attributes, providing context for developing predictive models for CRC recurrence risk factors and survival analysis, advanced machine learning algorithms will be utilized, including random forests, logistic regression, support vector machines, and decision trees. The effectiveness of these models in predicting CRC recurrence will be evaluated using ROC curve analysis. To mitigate overfitting and enhance the robustness of the results, cross-validation will be systematically conducted on the training dataset. Feature selection will be performed using Sequential Backward Search Selection, initially considering all features and progressively eliminating non-informative ones in each iteration to enhance model performance. The performance of the machine learning models will be assessed using various metrics.