The use of modified cellulose beads to remove zinc (II) from aqueous solution via a column study. A fixed-bed column was used to remove zinc (II) from aqueous solution at room temperature, and the evaluation process factors included concentration (5–15 mg/L), flow rate (5–9 mL/min), and bed height (7–13 cm). The results confirmed that breakthrough occurred more quickly for lower bed heights, greater flow rates, and higher zinc (II) concentrations, with the Bohart–Adams and Thomas models proving to be the most appropriate kinetic models. Deep learning models, such as adaptive neuro-fuzzy inference systems and artificial neural network models with three algorithms, namely, purelin, logsig, and transig, were effectively used to model the effectiveness of zinc (II) removal in aqueous solutions via cellulose nanocrystal beads. Statistical approaches such as the Marquardt percentage standard deviation (MPSD), root mean square error (RMSE), mean square error (MSE), and coefficient of determination (R2) were used to compare the predicted results against the experimental data. The artificial neural network (ANN) model indicated a strong correlation value of 0.998; thus, to a greater extent, the trained models could effectively predict the adsorption process for zinc (II) from aqueous solution.