COMBINI: connecting COmplementary Medicine evidence and BIological kNowledge to support Integrative Health - Project Summary Complementary medicine (CM) approaches are increasingly used by health care consumers, accepted by the medical community, and viewed as a cornerstone of whole person health. However, much about the effectiveness and safety of CM approaches, as well as the mechanisms through which they affect human health and well-being, remain poorly understood. Published literature is a growing source of evidence on CM approaches, their effect on human health, and their biological mechanisms. However, much of this evidence remains in unstructured text and specialty journals. Furthermore, the quality of this evidence is often questioned. The size, growth, and the quality of the literature makes it difficult for researchers and clinicians to access reliable evidence on these topics. Concurrently, the number of curated databases on CM is growing, but they remain limited to relatively narrow subtopics. Comprehensive resources and tools focusing on CM approaches are currently lacking. For systematic use of the high-quality evidence on these topics for medical discovery and patient care and effective integration of CM approaches with conventional medicine, scalable methods to distill, standardize, and aggregate knowledge from disparate research literatures (e.g., CM, human metabolism, microbiome, immunology) and curated databases are needed. We hypothesize that informatics approaches, in particular natural language processing (NLP) combined with ontologies and knowledge graphs (KGs), can underpin such consolidation and integration. In this project, we aim to develop and validate comprehensive knowledge resources and NLP methods for mining the literature on CM approaches including their mechanisms of biological action (which we dub COMB literature). We will integrate the mined information with knowledge from curated databases in a KG to support knowledge management and hypothesis generation applications. Specifically, we aim to: (1) construct informatics resources to support COMB-related knowledge integration and extraction; (2) develop NLP methods to mine COMB knowledge from biomedical literature; (3) construct a COMB knowledge graph from literature and curated databases and demonstrate its utility for question answering and hypothesis generation. The successful completion of this project will deliver a comprehensive ontology of CM interventions and their biological mechanisms, the first annotated corpus broadly focusing on CM approaches, novel NLP models, and an integrative KG on CM approaches and their effects on human health. Furthermore, validation of these resources and tools on real- world CM questions by domain experts will demonstrate their potential for patient care and scientific discovery. We anticipate that the KG can be integrated with other biomedical knowledge bases and with evidence generated in omics studies (e.g., metagenomics, metabolomics) as well as clinical data (e.g., electronic health records), bringing us closer to a more complete understanding of the mechanisms involved in whole person health and improving human health and well-being.
