Optimization of gelatin methacryloyl hydrogel properties through an artificial neural network model

Abstract

Gelatin methacryloyl (GelMA) hydrogels are promising materials for tissue engineering applications due to their biocompatibility and tunable properties. However, the time-consuming process of preparing GelMA hydrogels with desirable properties for specific biomedical applications limits their clinical use. Visible-light-induced cross-linking is a well-known method for the preparation of GelMA hydrogels; however, a comprehensive investigation on the influence of critical parameters such as Eosin Y (EY), triethanolamine (TEA), and N-vinyl-2-pyrrolidone (NVP) concentrations on the stiffness and gelation time has yet to be performed. In this study, we systematically investigated the effect of these critical parameters on the stiffness and gelation time of GelMA hydrogels. We developed an artificial neural network (ANN) model with three input variables, EY, TEA, and NVP concentrations, and two output variables, Young's modulus and gelation time, derived from our experimental design. Through the alteration of individual chemical concentrations, [EY] between 0.005 and 0.5 mM and [TEA] and [NVP] between 10 and 1000 mM, we studied the impact of these alterations on the real-time values of stiffness and gelation time. Furthermore, we demonstrated the validity of the ANN model in predicting the properties of GelMA hydrogels. We also studied cell survival to establish nontoxic concentration ranges for each component, enabling safer use of GelMA hydrogels in relevant biomedical applications. Our results showed that the ANN model can accurately predict the properties of GelMA hydrogels, allowing for the synthesis of hydrogels with desirable stiffness for various biomedical applications. In conclusion, our study provides a comprehensive library that characterizes the stiffness and gelation time and demonstrates the potential of the ANN model to predict these properties of GelMA hydrogels depending on the critical parameters. The ANN models developed here can facilitate the optimization of GelMA hydrogels with the most efficient mechanical properties that resemble a native extracellular matrix and better address the need in the in vivo microenvironment. The approach of this study is to bring research about the synthesis of GelMA hydrogels to a new level where the synthesis of these hydrogels can be standardized with minimum cost and effort.