Project description:A leading cause of global mortality is coronary artery disease. Unfortunately, in-stent restenosis and thrombosis often compromise the long-term efficacy of percutaneous coronary intervention treatments. Current metallic and polymeric stents fundamentally fail to integrate with the vessel wall, instead triggering pathological inflammatory and fibrotic responses. To address these problems, we turned to a literature-informed generative AI-driven platform to design and validate a self-expanding, bio-instructive endovascular liner that effectively prevents restenosis. Using this platform, we identified an optimal hydrogel formulation that combines native-vessel compliance and fatigue resistance with anticoagulant activity, which we report here. In both rabbit abdominal aorta and preclinical porcine coronary artery models, the hydrogel liner demonstrated superior long-term vessel patency and minimal neointimal hyperplasia compared to clinical-grade bare-metal and bioresorbable stents. Spatial transcriptomic analysis revealed that the liner material actively reprograms the post-injury vascular microenvironment by downregulating genes associated with inflammation, fibrosis, and smooth muscle cell proliferation. This work establishes a new class of AI-designed vascular implants and represents a paradigm shift from mechanically passive or cytotoxic interventions to a bio-instructive strategy for suppressing pathological vascular remodeling.

2026-01-24 | GSE317119 | GEO

Assessing Generative Pretrained Transformers (GPT) in Clinical Decision-Making: Comparative Analysis of GPT-3.5 and GPT-4.

Project description: Not available

| S-EPMC11240076 | biostudies-literature