Machine Learning-Enhanced Multidisciplinary Assessment of Petroleum Hydrocarbon Pollution: Biochemical, Microbial, Toxicological, and Environmental Perspectives

Abstract

This paper presents a holistic analysis of the extent of petroleum-related pollutants and their toxicological implication in five Niger Delta Communities: Odimodi, Burutu, Obatebe, Ayakoromo, and Gbekebor. The hydrocarbon analyses showed that TPH and PAHs were repeatedly above the standards defined by the World Health Organization, with Gbekebor having the highest TPH (24.65 mg/L), and significant burdens of Pb (0.20 mg/L) and Cd (0.07 mg/L). Positive correlations between metals and hydrocarbons proved synergistic release due to changes in the redox conditions. Microbial tests indicated extremely high coliforms and E. coli, particularly with Burutu and Obatebe, where biochemical oxygen demand and oxygen depletion were considerable. Biomarker tests revealed dramatic physiological disturbances: the expression of CYP1A1 was shown to increase up to 4.8-fold, GST activity was significantly elevated, and hepatocellular stress was evident through increased ALT / AST ratios. Ecological indicators rank Odimodi and Burutu as high-risk areas, with values above 2.5 in Pollution Load Indices and the Hazard Quotient index for Pb and Cd in children exceeding 3, indicating very high levels of neurotoxic and nephrotoxic hazards. The RF and RF-ANN ensemble performers achieved successive ranks of 96.19%, 95.61%, and 95.91% in terms of predictive accuracy, with TPH, Pb, and Cd as the predominant pollution predictors. This paper integrates chemical, microbial, biomarker, and computational data to inform an authoritative risk assessment and a targeted environmental management solution for petroleum-polluted environments.