A Multi-scale Visual Analytics Approach for Exploring Biomedical Knowledge

Fahd Husain, Rosa Romero-Gómez, Emily Kuang, Dario Segura, Adamo Carolli Carolli, Lai Chung Liu, Manfred Cheung, Yohann Paris

External link (DOI)
View presentation:2021-10-24T14:10:00ZGMT-0600Change your timezone on the schedule page
2021-10-24T14:10:00Z
Exemplar figure, described by caption below
A biomedical researcher uses the graph prototype to investigate potential drug treatments for SARS-CoV-2 using the COVID-19 biological graph that was automatically derived from literature. (A) The Global View provides an overview of the biological graph with bundled edges and nodes organized hierarchically in a biomedical ontology. The results of searching for links from several articles using DOIs are highlighted. (B) The Local View shows the highlighted results extracted as a node-link flow graph for further analysis. (C) The Drill-down Panel displays the underlying evidence extracted from scientific articles for the inhibition relationship between tocilizumab and IL6.
Abstract

This paper describes an ongoing multi-scale visual analytics approach for exploring and analyzing biomedical knowledge at scale. We utilize global and local views, hierarchical and flow-based graph layouts, multi-faceted search, neighborhood recommendations, and document visualizations to help researchers interactively explore, query, and analyze biological graphs against the backdrop of biomedical knowledge. The generality of our approach - insofar as it re-quires only knowledge graphs linked to documents - means it can support a range of therapeutic use cases across different domains, from disease propagation to drug discovery. Early interactions with domain experts support our approach for use cases with graphs with over 40,000 nodes and 350,000 edges.