BioRAG AI Agent
Autonomously retrieves, classifies, embeds, and performs semantic search to extract structured data from PubMed for clinical trials.
소개
The BioRAG AI Agent is an intelligent system designed to automate the semantic retrieval and structured data extraction of clinical trial information from PubMed. It functions as a Retrieval-Augmented Generation (RAG) system, taking natural language prompts and converting them into optimized PubMed queries. Leveraging a Model Context Protocol (MCP) server for robust data fetching, the agent then classifies, embeds, and stores relevant articles in a FAISS vector database. Finally, it conducts semantic searches and utilizes large language models to extract key biomedical metadata such as drug names, indications, sponsors, and NCT IDs, providing a structured biomedical knowledge base.
주요 기능
- Converts natural language prompts into optimized PubMed queries using OpenAI GPT.
- Embeds selected records into a FAISS vector database for efficient semantic search.
- 0 GitHub stars
- Performs multi-step document classification using rule-based logic and dual LLMs (GPT-3.5-turbo, GPT-4o-mini).
- Autonomously retrieves and classifies PubMed articles via a rate-limited MCP server.
- Extracts structured biomedical entities (drug names, indications, sponsors, NCT IDs) from in-scope abstracts.
사용 사례
- Automated retrieval and extraction of clinical trial data from scientific literature.
- Semantic search and filtering of clinical trial documents based on specific criteria.
- Building structured biomedical knowledge bases from PubMed abstracts.
