ALT Figure 1 – Comparison of Baseline Large Language Model (LLM) and Retrieval Augmented Generation (RAG) LLM Model. The figure contrasts the workflows and outputs of a Baseline LLM Models such as ChatGPT with a Retrieval Augmented Generation (RAG) LLM Model. (A) The Baseline LLM Model shows a direct pathway where a user query is processed by the LLM to generate a respon se. (B) Our RAG LLM Model enhances this process by incorporating a search for relevant information from a knowledge database containing over 200,000 abstracts from key vascular surgery journals, latest clinical guidelines, and clinical trials. The user que ry combined with enhanced context is then processed by the LLM to produce an improved response. Additionally, variables such as prompt engineering and the quantity and quality of input/output across different versions of ChatGPT are adjusted to further refine the output.

ALT Collage of three male researchers. Two are wearing a suit and the other wearing clinical scrubs.