PubTator 3.0: an AI-powered literature resource for unlocking biomedical knowledge

Abstract

PubTator 3.0 (https://www.ncbi.nlm.nih.gov/research/pubtator3/) is a biomedical literature resource using state-of-the-art AI techniques to offer semantic and relation searches for key concepts like proteins, genetic variants, diseases and chemicals. It currently provides over one billion entity and relation annotations across approximately 36 million PubMed abstracts and 6 million full-text articles from the PMC open access subset, updated weekly. PubTator 3.0's online interface and API utilize these precomputed entity relations and synonyms to provide advanced search capabilities and enable large-scale analyses, streamlining many complex information needs. We showcase the retrieval quality of PubTator 3.0 using a series of entity pair queries, demonstrating that PubTator 3.0 retrieves a greater number of articles than either PubMed or Google Scholar, with higher precision in the top 20 results. We further show that integrating ChatGPT (GPT-4) with PubTator APIs dramatically improves the factuality and verifiability of its responses. In summary, PubTator 3.0 offers a comprehensive set of features and tools that allow researchers to navigate the ever-expanding wealth of biomedical literature, expediting research and unlocking valuable insights for scientific discovery.

Graphical Abstract

Figure 1.

PubTator 3.0 system overview and search results page: 1. Query auto-complete enhances search accuracy and synonym matching. 2. Natural language processing (NLP)-enhanced relevance: Search results are prioritized according to the strength of the relationship between the entities queried. 3. Users can further refine results with facet filters—section, journal and type. 4. Search results include highlighted entity snippets explaining relevance. 5. Histogram visualizes number of results by publication year. 6. Entity highlighting can be switched on or off according to user preference.

Figure 2.

(A) The PubTator 3.0 processing pipeline: AIONER (8) identifies six types of entities in PubMed abstracts and PMC-OA full-text articles. Entity annotations are associated with database identifiers by specialized mappers and BioREx (9) identifies relations between entities. Extracted data is stored in MongoDB and made searchable using Solr. (B) Entity recognition performance for each entity type compared with PubTator2 (also known as PubTatorCentral) (13) on the BioRED corpus (15). (C) Relation extraction performance compared with SemRep (11) and notable previous best systems (12,13) on the BioCreative V Chemical-Disease Relation (14) corpus. (D) Comparison of information retrieval for PubTator 3.0, PubMed, and Google Scholar for entity pair queries, with respect to total article count and top-20 article precision.