Data-driven predictive design of engineered living hydrogels
Engineered living materials (ELMs) offer a promising route to biologically manufactured materials for healthcare, construction and manufacturing. However, their rational design is limited by the lack of quantitative relationships linking design parameters to material properties. Here, we show that a tabular foundation model (TabPFN), informed by a small library of living hydrogels, accurately predicts macroscopic material properties from genetic and process parameters. Models were evaluated for predicting storage modulus (G'), fibre content, thickness, and permeability of Escherichia coli-produced living hydrogels containing CsgA-based fibres fused to genetically encoded PEG-like biopolymers. On an independent validation set, TabPFN achieved the strongest prediction for G' (R2 = 85.1%), reducing RMSE by 48.0% compared to linear regression. Property-guided design further enabled identification of parameters for achieving living hydrogels with desired properties. These results establish a broadly applicable framework for predictive design of ELMs, reducing experimental screening and accelerating the discovery of materials with targeted properties.