Excited to announce the project that @TiamFeridooni and I have been working on for the past year: an augmented LLM that combines the capabilities of open access models with publicly available vascular surgery resources for improved performance.
Our first paper with more to come!
Our first paper with more to come!
Our most recent study published in @JVascSurgVI demonstrating the profiency of vasc.ai in responding to VESAP-5 multiple choice qustionare.View quoted post on X
🔍 Study Highlights:
chat.vasc.ai performed with a 93.8% accuracy in answering complex, domain-specific questions, surpassing general AI models such as GPT-4o (77.7%).
This model minimizes the risk of misinformation, often seen in general-purpose AIs, by focusing on logical accuracy, making it a reliable tool for vascular surgery training, medical education, and patient care.
You can read the full study here:
jvsvi.org/article/S2949-…
For academic vascular surgeons, chat.vasc.ai presents a significant advancement in how specialized knowledge is curated and applied. Its potential to support clinical assessments, enhance educational resources, and integrate rapidly evolving research makes it a valuable tool in both academic and clinical settings.
#VascularSurgery #SurgicalEducation #RAG #AI #LLM