Artificial intelligence is transforming bioengineering by enabling data-driven discovery and optimization of therapeutic biomaterials. However, the predictive capability of AI ultimately depends on experimentally validated quantitative frameworks that establish fundamental relationships between molecular structure, material properties, and biological function. Here, we present an integrated bioengineering framework that bridges this critical gap and provides the quantitative foundation for the next generation of AI-guided therapeutic biomaterial design.
Using the clinically relevant dual glucagon-like peptide-1/glucagon receptor agonists oxyntomodulin (Oxm) and Aib2-oxyntomodulin (Aib2-Oxm) as model systems, we integrated cryogenic electron microscopy, atomic force microscopy, quartz crystal microbalance with dissipation monitoring, circular dichroism spectroscopy, and quantitative physical modelling to establish direct relationships between supramolecular architecture, nanomechanics, thermodynamic stability, fibril assembly kinetics, and controlled therapeutic release. We identify previously unrecognized ribbon-like hierarchical fibrillar architectures, quantify mechanical properties spanning the gigapascal regime, and establish the first integrated quantitative framework describing fibril elongation, activation energetics, equilibrium stability, and temperature-dependent peptide dissociation. Together, these results demonstrate how subtle molecular modifications propagate across multiple length scales to determine biomaterial mechanics, stability, and sustained therapeutic performance.
Beyond advancing long-acting peptide therapeutics, this work introduces a generalizable paradigm for AI-enabled biomaterial engineering. By integrating structural, mechanical, thermodynamic, and kinetic descriptors within a unified experimentally validated framework, it establishes physically interpretable features for future machine-learning models capable of predicting biomaterial behaviour and accelerating the rational design of next-generation therapeutic systems. This study demonstrates that rigorous quantitative bioengineering provides the essential experimental foundation upon which trustworthy AI-driven biomedical innovation can be built.