Interpretable AI-driven materiomics to decode microenvironmental cues for stem cell immunomodulation

Hydrogels that mimic the extracellular matrix create microenvironments containing diverse physicochemical cues that regulate stem cell fate, particularly their immunomodulatory and pro-regenerative functions. However, elucidating how microenvironmental cues regulate cell fate remains challenging because of complex multi-parameter interactions and nonlinear relationships, thereby requiring large numbers of samples and substantial experimental effort. Here, we present a materiomics platform that integrates a large-scale systematic dataset with artificial intelligence (AI) to evaluate and predict optimal stem cell niche features for enhancing the immunomodulatory functions of mesenchymal stem cells (MSCs). Specifically, binary hydrogels composed of methacrylated alginate and gelatin (AMGM) were used to construct a comprehensive materiomics database incorporating seven critical physicochemical cues associated with MSC-mediated regulation of macrophage polarization. Interpretable machine learning models trained on this materiomics database successfully predicted hydrogel formulations that enhanced MSC immunomodulatory efficacy and further identified matrix stiffness as the physicochemical cue with the greatest contribution to MSC-mediated immunomodulation within the AMGM hydrogel system. In summary, this study demonstrates the utility of AI for evaluating the relative contributions of microenvironmental cues and establishes a data-driven research framework for investigating microenvironmental regulation of stem cell function.