Pub-Guard-LLM is the first specialized AI system designed to detect fraudulent practices in biomedical scientific publications while providing reliable explanations for its assessments.

• Offers three deployment modes: vanilla reasoning, retrieval-augmented generation, and multi-agent debate
• Achieves superior performance compared to general-purpose LLMs in identifying fraudulent biomedical research
• Provides transparent reasoning to support detection claims, enhancing trust and adoption
• Creates a specialized benchmark from PubMed retractions to evaluate fraud detection capabilities

This research addresses the growing crisis of fraudulent publications in medicine that threatens patient safety, research integrity, and resource allocation. By automating detection with reliable explanations, Pub-Guard-LLM helps protect the scientific knowledge base that informs clinical practice.

Pub-Guard-LLM: Detecting Fraudulent Biomedical Articles with Reliable Explanations