The architecture of a tissue-engineering scaffold affects how cells attach, migrate, and remodel newly formed tissue. At present, many scaffolds are still developed through repeated adjustment of a few structural variables, making it difficult to balance porosity with mechanical strength or degradation with the pace of tissue formation. Generative artificial intelligence may help shift scaffold development toward a target-driven process. Given a desired biological or mechanical outcome, a model could identify structures that are likely to achieve it. This possibility depends on training data that connect scaffold features with biological responses. In this review, we consider the use of generative methods together with bioinformatics for scaffold design. We describe potentially useful data sources, including micro-computed tomography (CT) images, mechanical measurements, single-cell RNA sequencing, and degradation data. We also examine generative adversarial networks, variational autoencoders, diffusion models, and evolutionary algorithms. Rather than treating these approaches as interchangeable, we discuss the design tasks for which their strengths differ. Examples from cardiac, bone, cartilage, and related applications show how tissue context shapes data requirements and model selection. We close by considering data comparability, validation across length scales, and the practical challenges of clinical use. These issues will determine whether generative tools can support more tailored scaffold design.
Deng et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: