Reproduce “PriHA: A RAG-Enhanced LLM Framework for Primary Healthcare Assistant in Hong Kong” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Richard Wai Cheung Chan, Shanru Lin, Ya-nan Ma, Hao Chen, Liangjun Jiang, Wenqi Fan

Researchers from The Hong Kong Polytechnic University developed PriHA, a RAG-enhanced LLM framework designed as a primary healthcare assistant for Hong Kong, addressing the challenges of fragmented, inaccurate, and outdated health information. The system achieved an overall mean score of 4.197 across five evaluation metrics on a custom HK-PriHCQA dataset, surpassing zero-shot LLMs and single-source RAG approaches by integrating curated local knowledge with real-time web retrieval and an intelligent query optimizer.

Key insights

  • Combining a curated local knowledge base with a controlled, real-time web search agent is essential for RAG systems to provide both accurate and up-to-date information in dynamic, localized, and high-stakes domains like healthcare.
  • A sophisticated reconciliation module that prioritizes authoritative and time-valid content from diverse sources is crucial for preventing contradictory statements and enhancing trustworthiness in AI-generated health advice.
  • Implementing a proactive, multi-turn query optimizer, guided by domain-specific principles like a 'community-first' approach, significantly improves user intent understanding and guides users toward relevant primary care resources.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.