Integrated Survey and Clinical Data for Risk Assessment and Survival Prediction in Colorectal Cancer Using Machine Learning

Colorectal cancer (CRC) continues to be a significant contributor to global mortality rates, with its incidence influenced by various factors including dietary habits, genetic predispositions, and age demographics. This study uses a dual-dataset strategy that combines survey and clinical data, providing a unique combination of laboratory-validated clinical information and patient-reported outcomes to improve the accuracy of CRC survival forecasts. Machine learning classifiers were utilized to assess each dataset independently and to categorize individuals into high-risk and low-risk survival groups based on age and gender. Key variables, including treatment type, survival status, and disease-free survival (DFS) events, were analyzed across the datasets, uncovering significant similarities within the respective demographic groups. The Random Forest algorithm attained the highest accuracy of 93 % on the survey data, while the logistic regression model reached an accuracy of 96% on the clinical dataset compared to existing work. Our work suggests that patients under 50 and over 70 face an increased risk, while patients between 50 and 70 exhibit a comparatively lower risk when they receive prompt treatment. By incorporating both clinical and demographic variables in the prediction of colorectal cancer survival, machine learning models prove to be effective in recognizing high-risk profiles and guiding personalized treatment plans. This technique improves the accuracy of predictions regarding colorectal cancer outcomes, serving as a crucial resource for targeted interventions and better patient prognoses.