Active Learning-Accelerated Discovery of Fibrous Hydrogels with Tissue-Mimetic Viscoelasticity

The extracellular matrix (ECM) exhibits tissue-specific viscoelasticity with a unique combination of elasticity and viscosity. Hydrogels with independently controlled elastic modulus and stress relaxation have been developed. Yet, time- and labor-efficient identification of formulations used for the preparation of multicomponent hydrogels recapitulating the mechanical properties of a specific tissue remains a challenge. Conventional bottom-up screening of hydrogel formulations is resource-intensive, especially for navigating multicomponent hydrogels. Here, we present an active learning framework based on multi-objective Bayesian optimization to accelerate the discovery of biomimetic fibrous hydrogels replicating the elastic and viscous properties of the ECM of healthy and diseased tissues. In the data-driven approach, we iteratively navigated the multidimensional chemical space, while simultaneously optimizing hydrogel’s conflicting mechanical properties and mapping the boundaries of material feasibility using sparse experimental feedback. The experimental data were used to train a digital twin model and unveil the interplay between the fabrication conditions and hydrogel properties. The decoupled effects of the elasticity and viscosity were established for cell activation, proliferation, and morphogenesis. This work shows the potential of using the data-driven approach, leveraging even sparse experimental data to engineer hydrogels and optimize diverse biomaterials for tissue engineering and regenerative medicine.

The hydrogel components, that is, a-CNCs and GelMA were synthesized as described elsewhere [34]. GelMA-L and GelMA-H have an averaged degree of modification with methacryloyl groups of 25% and 75%, respectively. The stock of 3.0–6.0 wt.% suspension of a-CNCs was prepared in 1X Hank’s buffered salted solutions (HBSS). The stock solutions of GelMA-L, GelMA-H, and LAP were prepared at concentrations of 15.0, 15.0, and 2 wt.% in 1X HBSS, respectively. The hydrogel precursor was prepared using the concentrations recommended by the AL algorithm. Gelation was induced by incubating the mixtures of a-CNCs, GelMA-L, GelMA-H, and LAP for 2 h at 37°C, followed by UV irradiation for the ML-recommended duration (400 nm, intensity at 30 mW cm−2).

The disk-shaped samples (5.00 mm in height and 9.83 mm in diameter) for compression tests were prepared by crosslinking the hydrogel precursor mixture in 2 mL plastic syringes (Norm-Ject) with the tip cut off. Compression tests were performed using a Mach-1 Mechanical tester (Biomomentum Inc.). The samples were subjected to 20% compressive strain at a rate of 0.1 mm s−1. The Young’s modulus was determined from the slope of the linear region of the stress-strain curve at a 10%–15% strain.

To measure stress relaxation, the hydrogel precursor mixture was gelled between the 40 mm parallel plates of the DHR-1 rheometer (TA Instruments). Mineral oil was added to the edge of the mixture to prevent solvent evaporation. The sample was UV irradiated from underneath for a specific time, a 5% step strain was applied for 0.1 s and subsequently, held for 100 s. The relaxation modulus was calculated by dividing the measured shear stress at each time point by the applied strain [66]. This value of the relaxation modulus was normalized to its value at 0.1 s. Stress relaxation, σ100 was expressed as the relative decrease (in %) of the normalized relaxation modulus 0.1 s.

Cancer, fibrosis, hydrogels, machine learning, tissue scaffolds & viscoelasticity.