Artificial Intelligence tool for prediction of ECM mimics hydrogel formulations via click chemistry

Abstract

A user-friendly machine learning (ML) predictive tool is reported for designing extracellular matrix (ECM)-mimetic hydrogels with tailored rheological properties. Developed for regenerative medicine and 3D bioprinting, the model leverages click chemistry crosslinking to fine-tune the mechanical behaviour of gelatin- and hyaluronic acid-based hydrogels. Using both experimental rheological data and synthetic datasets, our supervised ML approach accurately predicts hydrogel compositions, significantly reducing the cost and time associated with trial-and-error approach. Despite advancements in the field, existing models remain limited in their ability to mimic the ECM due to the use of non-natural polymers, reliance on a single type of biologically active macromolecule, and physical crosslinking reactions with limited tuneability. Additionally, their lack of generalizability confines them to specific formulations and demands extensive experimental data for training. This predictive platform represents a major advancement in biomaterial design, improving reproducibility, scalability, and efficiency. By integrating rational design, it accelerates tissue engineering research and expands access to customized ECM-mimetic hydrogels with tailored viscoelastic properties for biomedical applications, enabling both experts and non-experts in materials design. • Machine learning tool guiding the synthesis of ECM-mimics with customized rheological properties. • Gelatin-hyaluronic acid hydrogels crosslinked with PEG star via click chemistry. • Biomaterial design accelerated by machine learning powered by experimental and synthetic data. • A.I. tool for ECM-mimics design, allowing customizable generation of bioactive hydrogels with improved functionality. • Platform enhancing reproducibility, scalability, and accessibility, in advancing tissue engineering and biomaterial research.

Topics & Concepts