AI-Guided Biosorption of Lead Using Prosopis cineraria Leaf Powder in an Omani Industrial Wastewater Matrix: Kinetics, Thermodynamic Trends, and Predictive Modeling

The present study introduces a hybrid experimental-computational framework, which combines experimental biosorption with artificial neural network (ANN) to support the data-driven optimization of Pb(II) extraction from Omani industrial wastewater. A sustainable biosorbent derived from Prosopis cineraria leaves, referred to as P. cineraria leaf powder, is a locally grown and unmodified biomass that was considered as a batch-based biosorbent. Systematic investigation of the operational parameters such as pH, contact time, biosorbent dosage, particle size, and initial Pb(II) concentration was conducted. Under slightly acidic conditions, rapid uptake occurred, and the equilibrium reached within 30 min and adsorption capacity was 31.97 mg/g. Kinetic analysis was well described by the pseudo-second-order model ( R 2 = 0.999), while Fourier transform infrared spectroscopy analysis indicated the involvement of hydroxyl and carboxyl functional groups in Pb(II) uptake; X-ray diffraction and scanning electron microscopy further demonstrated changes in crystallinity and surface morphology after adsorption. Observed thermodynamic trends were apparent indicators of spontaneous and endothermic uptake of Pb(II) throughout the temperature range (303–318 K) with the ability to interpret at the level of the system. A 3-6-1 ANN model showed a high level of predictive accuracy ( R 2 = 0.999; RMSE = 1.18; MAE = 0.82) that exceeded that of traditional regression-based models and allowed a model to be quickly used to predict removal efficiency in different operating conditions. Overall, the proposed AI-assisted biosorption approach demonstrates the potential of integrating locally available plant-based biomass with data-driven modeling for Pb(II) removal from industrial wastewater and provides a promising basis for further investigation of its practical applicability.