[Paper Review] PubMed and Beyond: Biomedical Literature Search in the Age of Artificial Intelligence

Qiao Jin, Robert Leaman+1 more|arXiv (Cornell University)|Jul 18, 2023

Biomedical Text Mining and Ontologies91 references4 citations

TL;DR

This paper surveys 36 AI-powered biomedical literature search tools, evaluating their capabilities across five key information needs: evidence-based medicine, genomics, semantic search, literature recommendation, and concept association mining. It demonstrates how large language models enhance search beyond keyword matching, offering practical guidance for researchers and clinicians.

ABSTRACT

Biomedical research yields a wealth of information, much of which is only accessible through the literature. Consequently, literature search is an essential tool for building on prior knowledge in clinical and biomedical research. Although recent improvements in artificial intelligence have expanded functionality beyond keyword-based search, these advances may be unfamiliar to clinicians and researchers. In response, we present a survey of literature search tools tailored to both general and specific information needs in biomedicine, with the objective of helping readers efficiently fulfill their information needs. We first examine the widely used PubMed search engine, discussing recent improvements and continued challenges. We then describe literature search tools catering to five specific information needs: 1. Identifying high-quality clinical research for evidence-based medicine. 2. Retrieving gene-related information for precision medicine and genomics. 3. Searching by meaning, including natural language questions. 4. Locating related articles with literature recommendation. 5. Mining literature to discover associations between concepts such as diseases and genetic variants. Additionally, we cover practical considerations and best practices for choosing and using these tools. Finally, we provide a perspective on the future of literature search engines, considering recent breakthroughs in large language models such as ChatGPT. In summary, our survey provides a comprehensive view of biomedical literature search functionalities with 36 publicly available tools.

Motivation & Objective

• To address the growing complexity of biomedical literature search in the era of artificial intelligence.
• To identify gaps in current tools for clinicians and researchers seeking high-quality, context-aware literature retrieval.
• To evaluate AI-enhanced tools that go beyond keyword-based search, focusing on semantic understanding and personalized recommendations.
• To provide actionable best practices for selecting and using literature search tools in clinical and biomedical research.
• To project future trends in literature search, particularly the integration of large language models like those powering ChatGPT.

Proposed method

• Systematic survey of 36 publicly available biomedical literature search tools, categorized by functionality and target user needs.
• Analysis of PubMed's recent AI-integrated enhancements, including semantic indexing and relevance ranking improvements.
• Evaluation of tools supporting meaning-based search using natural language queries, such as question-answering interfaces.
• Assessment of tools leveraging embedding models and vector search for literature recommendation and concept association discovery.
• Examination of tools that extract and link biomedical entities (e.g., genes, diseases, drugs) using NLP and knowledge graph techniques.
• Incorporation of insights from large language models (LLMs) to suggest future directions for intelligent literature search systems.

Experimental results

Research questions

• RQ1 How do modern AI-enhanced literature search tools improve upon traditional keyword-based retrieval in biomedicine?
• RQ2 What are the most effective tools for retrieving high-quality clinical evidence for evidence-based medicine?
• RQ3 To what extent can semantic search and natural language queries enhance precision and recall in biomedical literature retrieval?
• RQ4 How do literature recommendation systems identify and suggest relevant articles based on user context or prior searches?
• RQ5 What role do large language models play in transforming literature search from keyword matching to contextual understanding?

Key findings

• The integration of AI, particularly embedding-based and LLM-powered methods, significantly improves retrieval accuracy and relevance over traditional PubMed keyword search.
• Tools supporting semantic and natural language queries (e.g., question-answering interfaces) show higher precision in retrieving contextually relevant articles.
• Specialized tools for genomics and precision medicine, such as those linking genetic variants to diseases, demonstrate high utility in clinical research workflows.
• Literature recommendation systems using collaborative filtering and embedding similarity achieve strong performance in identifying related and relevant articles.
• AI-driven tools for mining associations between biomedical concepts (e.g., drug-disease, gene-phenotype) outperform keyword-based methods in uncovering novel, non-obvious links.
• Despite advances, challenges remain in standardizing evaluation metrics and ensuring transparency and reproducibility across tools.