Jeannie Yuhaniak Irwin

ONYX: A System for the Semantic Analysis of Clinical Text

This paper introduces ONYX, a sentence-level text analyzer that implements a number of innovative ideas in syntactic and semantic analysis. ONYX is being developed as part of a project that seeks to translate spoken dental examinations directly into chartable findings. ONYX integrates syntax and semantics to a high degree. It interprets sentences using a combination of probabilistic classifiers, graphical unification, and semantically annotated grammar rules. In this preliminary evaluation, ONYX shows inter-annotator agreement scores with humans of 86% for assigning semantic types to relevant words, 80% for inferring relevant concepts from words, and 76% for identifying relations between concepts.

The dental informatics online community.

Dental Informatics (DI) is the application of computer and information science to improve dental practice, research, education, and program administration. To support the growth of this emerging discipline, we created the Dental Informatics Online Community (DIOC). The DIOC provides a dedicated professional home for DI researchers and serves as an open, common, and worldwide forum for all individuals interested in the field. It was created and is maintained by the Center for Dental Informatics at the University of Pittsburgh School of Dental Medicine, independent from any professional association, corporate interest or funding source. The DIOCs Website provides many useful features, such as a learning center, publication archive, member and project directories, and the Current Awareness Service (CAS). The CAS automatically notifies members about new information added to the Community. Notifications are individualized according to a members profile and activities on the site. The DIOC is a research-oriented online community which provides resources in the dental informatics and dental technology field, as well as a way to establish social connections to share ideas, problems and research opportunities. Member and activity growth since the communitys inception in 2005 have been steady, but future sustainability of the community depends on many factors.

PS2-37: Oral Cancer Information on the Web: Assessing the Quality and Content of English and Spanish Oral Cancer Websites

Background/Aims: In the United States, about 8,000 people a year die from oral cancer and more than 30,000 new cases are diagnosed annually. A recent study showed that 80% of American adult Internet users have searched the Web for health information and 15% of those specifically searched for dental health information. Having high quality oral cancer information available via the Web is important given the significance of this health problem. The goal of this study was to evaluate the quality and content of multiple English and Spanish oral cancer websites. Methods: We developed a search strategy using the keywords: oral cancer, mouth cancer, and tongue cancer to find oral cancer sites via Medline Plus, Google, and Yahoo. We then used the translations cancer oral, cancer de la boca, and cancer de la lengua to search Medline Plus en Español, Google Español, and Yahoo Telemundo. We added sites to the datasets based on inclusion/exclusion criteria. Two native speaking raters evaluated each site within their set for quality using the modified Information Quality Tool (IQT). We then developed a survey tool to asses the content of the sites. Two native speaking oral cancer experts evaluated each site within their set using this new tool. Results: Our search strategy produced 24 English language sites and 24 Spanish language sites for evaluation. English language websites had an average IQT score of 74.7 (out of 100) and average content score of 51.5 (out of 100). Spanish-language sites had an average IQT score of 48.8 and an average content score of 25.9. Conclusions: Despite higher scores for the English language websites, our analysis showed that there was a great variation in overall quality and content with room for improvement for both language types. English sites could make the biggest improvements by providing more information about their sponsors and who controls site content as well as updated and fixing links and author credentials. The Spanish sites should focus on improving information about their agenda and overall site design. For content, both types of sites need to provide more images and information on etiology and risk factors. Both English and Spanish sites scored highest in coverage of clinical presentation, treatment and follow-up care after treatment. Overall, this study identifies several information gaps and areas for improvements for oral cancer websites. We hope our results will be used to improve online oral cancer information.