Abstract Accurate identification of bladder cancer is critical for optimizing patient management. This study aimed to develop a machine learning–based case–control classification model that integrates clinical data, cytokine levels, and environmental factors to discriminate bladder cancer cases from benign controls in an Iraqi population, complemented by Cox proportional hazards analyses of survival outcomes among diagnosed patients. A retrospective, multicenter cohort of 600 bladder cancer patients and 500 benign bladder tumor controls was analyzed. Machine learning models including Extreme Gradient Boosting, Logistic Regression, and Support Vector Machine were trained using 15 clinical variables, plasma cytokine levels (IL-6, IL-7, IL-15, IL-33), immunohistochemical tissue expression, and nine environmental risk factors. Model performances were compared to identify the most predictive tool. During an 11-year median follow-up, 19.7% of patients died. The Extreme Gradient Boosting model demonstrated superior predictive performance, achieving 85.2% accuracy and an area under the curve of 0.924. Key predictors identified included smoking, age, dietary habits, liver function, water sources, diabetes, and alcohol consumption. In the multivariate Cox analysis, sex remained the only variable with a statistically significant association with survival outcomes. This machine learning–based model provides an effective approach to classify bladder cancer cases versus benign controls in Iraq, highlighting the integrated contribution of clinical, environmental, and cytokine features, and providing a foundation for future studies in larger and independent cohorts.
ALshadood et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: