Integrative Approach: Yoga-Based Prevention and Venous Chronic Severity Score Control and Chronic Venous Disease Enhanced by Machine Learning Predictive Models

Chronic venous disease (CVD), a common vascular disorder caused by impaired venous return, results in pain, swelling, skin changes, and reduced quality of life. Prolonged standing, obesity, and inactivity accelerate disease progression. The Venous Clinical Severity Score (VCSS) is widely used in clinics as an indicator to assess the severity of the disease and monitor the patient's response over time. Recently, yoga has been gaining popularity as a complementary therapy to improve blood flow as well as to help in the management of symptoms in patients with venous disorders. Using demographic, lifestyle, vascular, clinical, and quality-of-life-related factors linked to CVD, this study suggests a machine learning (ML)-based prediction methodology for VCSS severity evaluation. Two widely used classification algorithms, namely decision tree classification and support vector classification (SVC), were employed and subsequently enhanced using the Runge-Kutta optimization and Rhizotomy optimization algorithms. The resulting hybrid models, DTRO, DTRK, SVRO, and SVRK, were developed and evaluated for VCSS severity prediction. Exploratory data analysis, data balancing, and preprocessing were carried out before modeling to make the model more robust. The model's performance was evaluated by multiple assessing measures such as accuracy, precision, recall, F1-score, confusion matrices, receiver operating characteristic (ROC) analysis, 5-fold cross-validation, and Wilcoxon signed-rank tests. Experimental results demonstrated that optimization-assisted models generally improved predictive performance compared with their corresponding baseline classifiers. In particular, the SVC-based hybrid models achieved strong and consistent classification capability across different VCSS severity categories, while ROC analysis yielded area under the curve values above 0.89 for all evaluated conditions. The improved models continuously showed competitive and dependable prediction performance, despite Wilcoxon analysis showing that performance differences were not statistically significant at the 95% confidence level. The findings suggest that optimization-assisted ML frameworks can effectively support VCSS severity prediction and may contribute to clinical risk stratification and decision support in CVD management.