Comparative Evaluation of Machine Learning-Assisted Statistical Modeling of Biopolymer Hydrogel Swelling for Material Optimization.

This study presents an in-depth examination of the swelling characteristics of biopolymeric hydrogels developed with sodium alginate, employing both conventional and modern modeling techniques to predict and optimize the process. Hydrogel swelling properties have found critical applications in controlled drug delivery, wound dressing, and food technology applications, and thus have developed a high demand for predictive and tuning strategies. In this case, different concentrations of polymers, cross-linkers, microalgal cells, and media volume are set with different compositional parameters. The measurements of experimental swelling degrees were conducted under controlled conditions to obtain both equilibrium and dynamic measurements. Initial data analysis was performed by response surface methodology (RSM) with a central composite design (CCD), which explained the effects of independent and interactive formulation factors. Then, an artificial neural network (ANN) model was implemented and trained on the same data set to elicit possible complex and nonlinear associations that are not considered in traditional approaches. The results of the two models were evaluated intensively based on the statistical parameters such as R