The rise in human activity has intensified environmental pollution, posing a global public health challenge. Understanding the intricate mechanisms by which pollutants impact health is crucial. Traditional research, often limited to specific techniques and short-term exposures, fails to capture the full complexity of these interactions. This study integrates machine-learning, quantum chemical computing, physicochemical properties, target prediction, KEGG and GO pathway analyses and survival analysis to examine the effects of air and water pollutants on human health. We chose knowledge-guided pre-trained graph transformer (KPGT) framework with an AUC of 0.83 knowledge bootstrap to predict the carcinogenic potential of pollutants and clustered environmental pollutants into seven different groups. For each group, quantum chemical and physicochemical properties, target prediction, KEGG and GO pathway analyses further revealed links between pollutants and cancer-related factors, such as MicroRNA, PD-L1, the PD-1 checkpoint pathway, and HIF-1 signaling. Survival analysis identified key proteins associated with poor cancer prognosis. These findings provide insights into the complex effects of pollutants on health and contribute to public health research.