William Hsu PhD, Presenter: Nothing to Disclose

Maurine Tong, Abstract Co-Author: Nothing to Disclose

Rick K. Taira PhD, Abstract Co-Author: Nothing to Disclose

Alex Anh-Tuan Bui MS, PhD, Abstract Co-Author: Nothing to Disclose

We present a framework for structuring, integrating, and visualizing scientific claims and associated context from biomedical papers. We demonstrate the utility of this information in translational research by facilitating hypothesis generation and knowledge discovery.

The rate of scientific discovery is greatly outpacing our ability to comprehend and apply this knowledge. Medline indexed over 2,900 papers about non-small cell lung cancer in 2012 alone. The number of papers reflects the fast pace at which insights are being generated. However, this growing body of literature also exposes our inability to effectively integrate and understand the vast body of evidence. Current literature retrieval tools do not provide functionality to easily identify and summarize studies based on participant demographics, study design, and measured variables. This information is useful in identifying areas of active research, studies reporting conflicting evidence, and gaps in our understanding of a disease. We have created an interactive visualization that permits clinical scientists with exploring information provided by biomedical papers, summarizing scientific claims, and understanding relationships among studies.

The visualization tool is presented in the context of characterizing the role of EGFR expression in treatment response of NSCLC patients through the lens of clinical, imaging, and genomic factors. Scientific claims from a set of 31 full-text papers were extracted, standardized, and mapped to entities and attributes in the standardized data model. Use cases were developed to guide development and usability assessment of the user interface.

This work addresses the need for a standardized data model for biomedical literature, text extraction tools to map information from full-text papers to the data model, and a web-based visualization to explore and query a large, multidimensional dataset. Informatics challenges related to the semantic characterization of scientific claims and integration of heterogeneous evidence encompassing multiple biological scales are discussed.

Hsu, W, Tong, M, Taira, R, Bui, A, Visualizing Biomedical Literature: Integration and Application of Clinical, Imaging, and Genomic Evidence Reported in Research Studies.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13044405.html