Antibiotic removal in South African water using artificial neural networks and adaptive neuro-fuzzy inference system models: A review

The growing occurrence of antibiotic residues in South African water systems poses serious environmental and public health risks, owing mostly to pharmaceutical discharge, agricultural runoff, and poor waste management. Conventional water treatment procedures frequently fail to properly remove these micropollutants, needing new predictive and analytical approaches. This review critically investigates the implementation of Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) models to forecast and optimize antibiotic removal from South African water bodies. To the best of our knowledge, little or no research compares the models' respective performances in the context of the urban water cycle in South Africa. Therefore, this review elaborates on some of the pharmaceuticals (such as diclofenac sodium and tetracycline) that have been studied, as well as the challenges associated with their removal. It also emphasizes studies on modeling and predicting pharmaceutical removal from wastewater using ANN and ANFIS models. Additionally, this review considered the comparisons between ANN and ANFIS models in predicting the removal of emerging contaminants, as well as the challenges and limitations associated with these modeling techniques. The studies established that AI models achieved higher R² and lower error metrics compared to classical statistical or isotherm models.