The talk is about automated generation of insights from evidence with a special focus on provenance.

The nearly universal adoption of electronic health records and rapid advancements in multiomics technologies have made tremendous amounts of data available for research. Unfortunately, these data are often highly distributed and heterogeneous which limits most researchers from utilizing them. Knowledge graphs (KGs) have frequently been used to systematically interrogate the biology underlying complicated systems and diseases. In this talk, I introduce novel methods designed to improve pediatric rare disease phenotyping.

Research progresses through accumulating knowledge such that a previously unexplored subject (an unknown unknown) becomes an active research area exploring the questions (known unknowns), until a body of established facts emerges (known knowns). Many knowledge-bases exists for known knowns, but no ignorance-bases exist for known unknowns. What novel connections and insights are in the unknowns? Using a knowledge graph, we created the first ignorance-base for prenatal nutrition to help find pertinent questions that could affect mothers and offspring globally.

Knowledge graphs can provide an ideal environment for representation and reasoning about the complexities of the patient, of the disease, and of clinical practice. It is critical, however, to incorporate temporal considerations of processes as well as the continuing evolution of data and and concepts, e.g., clinical protocols for diagnosis and treatment. The application of knowledge graphs as a "healthcare learning environment" that addresses health equity issues in infant/maternal morbidity and mortality as well as hypertensive disorders of pregnancy will be presented.

In the vast world of biology data, finding the right information can feel like searching for a needle in a haystack. This talk introduces an exciting solution: using AI chatbots paired with knowledge graphs. These tools can dive deep into large datasets, unveiling crucial insights that might otherwise stay hidden. We'll explore how this blend of technology helps, its safety benefits, and the ways it promises to revolutionize biology research. Imagine a powerful magnifying glass for data—this is the future of scientific research. Join us to see how AI is reshaping our understanding of the biological world. 

The Bridge2AI initiative, initiated by the National Institutes of Health (NIH), aims to broaden the application of artificial intelligence in both biomedical and behavioral research. This talk discusses a project we are engaged in that integrates knowledge graphs with quantum computing and machine learning, as a component of the NIH's Bridge2AI program.

Translational research is significantly enriched by public knowledge sources that provide annotations across various levels, such as drugs/treatments, diseases, targets, and metabolism. The value of these sources has surged in the last decade due to the broader availability and adoption of standardized identifiers, processes, and metadata capture. These enhancements have led to the creation of detailed graphs (networks) that amalgamate complementary data, offering new insights or revealing previously overlooked or unconsidered connections. Additionally, machine learning and artificial intelligence (AI/ML) methods can be utilized to predict outcomes, address a range of questions in translational research, and formulate innovative hypotheses. In this presentation, we highlight innovations within the National Center for Translational Sciences in generating graphs to identify repurposed drugs, novel biological mechanisms and to discover new associations between genes, targets and diseases.  Examples highlighted will include the NCATS Translator (https://ui.ci.transltr.io/demo) and the Rare Disease Alert System (https://rdas.ncats.nih.gov/) and more.