Groundwater contamination poses a significant public health risk, particularly in regions like Saudi Arabia where groundwater is a primary water source. This study evaluates the effectiveness of advanced machine learning (ML) models, Kernel Support Vector Machine (SVM-K), and Boosted Trees (BT), in predicting the Probability of Cancer Risk (PCR) from dermal intake of contaminated groundwater in Saudi Arabia. Using heavy metals like Arsenic (As), Chromium (Cr), Cadmium (Cd), and Lead (Pb) as predictors, the study compares two combinations of each model type, SVM-K (C1 and C2) and BT (C1 and C2), across several performance metrics. For SVM-k the MAPE values were extremely low, 0.1596% (C1) and 0.1429% (C2), with negligible negative PBIAS values of -0.0013 (C1) and -0.0045 (C2), suggesting highly accurate and unbiased predictions. In contrast, for BT, the MAPE values were higher at 0.9591% (C1) and 1.0214% (C2), with positive PBIAS values of 0.0401 (C1) and 0.0482 (C2), indicating a tendency to slightly overestimate PCR. The results emphasize the potential of SVM-K models in providing reliable and accurate PCR predictions, essential for effective public health interventions and groundwater management, aligning with global health and environmental sustainability goals. The integration of these findings could significantly improve the efficacy of public health advisories and targeted remediation efforts, thereby reducing the health impacts associated with contaminated groundwater.