Freeze-thaw-synthesized PVA/chitosan hydrogels: structure-property relationships and ANN modeling of swelling and degradation behaviors

Official URL: https://dx.doi.org/10.1021/acsomega.6c00485

Reliable prediction of swelling and degradation behavior of hydrogels used in biomedical applications due to the simultaneous and nonlinear effects of multiple parameters is a critical requirement in hydrogel design. However, conventional experimental approaches present significant limitations in multiparameter systems due to high time consumption and experimental costs. In this study, the structural, swelling, and degradation characteristics of poly(vinyl alcohol)/chitosan (PVA/CS) composite hydrogels synthesized using a physical cross-linking method based on freeze–thaw (F–T) cycles with varying chitosan contents and F–T cycle numbers were systematically investigated. To prioritize biocompatibility, the hydrogels were produced without the use of chemical cross-linking agents, and the effects of pH-sensitive behavior, network density, and structural stability were examined in detail. FT-IR analyses confirmed the formation of hydrogen bonding interactions and an IPN-like network structure between PVA and chitosan chains, while SEM observations revealed significant changes in pore morphology depending on chitosan content and the number of F–T cycles. The swelling capacities of the hydrogels were found to vary between 4.02 and 26.28 g water/g polymer, whereas degradation ratios ranged from 3.38% to 37.36%, depending on the environmental pH, chitosan concentration, and F–T cycle number. The experimentally obtained nonlinear swelling and degradation data were further modeled using an artificial neural network (ANN) approach, and the behaviors were predicted with high accuracy (R > 0.99). The results demonstrate that ANN-based modeling provides a reliable and efficient design and optimization tool for multiparameter hydrogel systems by significantly reducing experimental workload.