A deep learning technique (artificial neural network) was employed to predict the efficacy of biodegradable and environmentally friendly adsorbents. The temperature, Cd (II) concentration in water (mg/L), pH, contact duration (min), and amount of adsorbent used (mg/L) were considered. Two output variables exist: the adsorption capacity (mg/g) and the percentage removal. The thermal degradation of 90% occurred between 270 and 380 °C, with the carboxylate functional group being the primary component assisting in eliminating Cd (II). The carboxylate functional group facilitates the elimination of Cd (II). A maximum adsorption capacity of 270 mg/g with a percentage removal of 97% was achieved at pH 5, a contact time of 300 min, an adsorbent dosage of 20 mg, and a pollutant concentration of 315 mg/L. The Levenberg‒Marquardt algorithm was utilised to train the artificial neural network. It consists of 5 inputs, 2 outputs, and 10 hidden layers. The outputs of the ANN models were evaluated against real data via R2 and MSE. The ideal artificial neural network (ANN) model produced both accurate and dependable findings, with R2 values of 0.997 and an error of 0.014. The Langmuir model most accurately explained the adsorption of Cd (II). The model showed very high accuracy, with an R2 value of 0.997 and very small errors, including a mean percentage standard deviation (MPSD) of 0.0012, a root mean square error (RMSE) of 0.028, and a mean squared error (MSE) of 0.008. The pseudo-second-order model describes the adsorption process, whereas the Dubinin‒Radushkevich model suggests that the process is guided by a chemical reaction.