Advances in translational research has led to the need for well annotated biospecimen for research. The Mesothelioma Virtual Bank (MVB, http;//www.mesotissue.org) is one such initiative to gather clinically annotated datasets relevant to human mesothelioma to develop an enterprise biorepository for researchers. The MVB architecture is based on three major components: (a) common data element (based on CAP protocol and NAACCR standards), (b) clinical and epidemiologic data annotation, and (c) data query tools. These tools work interoperably to rationalize the entire process of annotation. The MVB tool is based upon the caTISSUE Clinical Annotation Engine, developed by the University of Pittsburgh in cooperation with the Cancer Biomedical Informatics Grid (caBIG, see http://cabig.nci.nih.gov). This application provides a web-based system for annotation, importing and searching mesothelioma cases. Underlying information model is constructed utilizing Unified Modeling Language (UML) class diagrams, hierarchical relationships and Enterprise Architect (EA) software. The database provides researchers real-time access to a richly annotated specimen and integral information related to mesotheliomas. The data disclosed are tightly regulated depending upon users’ authorization and depending on the participating institute is amenable to the local IRB and regulation committee reviews. The MVB currently has over 400 annotated cases available for researchers that include paraffin embedded tissues, tissue microarrays, serum and genomic DNA. The Mesothelioma Virtual Bank (MVB) is a virtual biospecimen registry with robust translational biomedical informatics support to facilitate basic science, clinical, and translational research. Furthermore, it protects patient privacy by disclosing only de-identified data and biospecimens can be efficiently made accessible to researchers.

Extending Text Categorization Filters in Medicine
Yin Aphinyanaphongs, Vanderbilt University, Nashville, TN, Nancy Wilczynski, McMaster University, Hamilton, Ontario, Canada, Constantin Aliferis, Vanderbilt University, Nashville, TN, and Brian Haynes, McMaster University, Hamilton, Ontario, Canada
Monday, November 12, 10:30 am - 12:00 pm
Location: Chicago 9

In previous studies, we explored the potential of machine learning methods for biomedical information retrieval in support of Evidence-Based Medicine. We specifically developed machine learning filter models that can identify high quality articles in internal medicine in the treatment, prognosis, diagnosis, and etiology content categories with high predictivity. In the present study, we extend the previous work by building and evaluating machine learning filter models that identify high quality articles in areas outside of internal medicine and in several format, purpose, and rigor content categories. We used a gold standard encompassing 49,028 articles in 161 journals in areas including pediatrics, psychology, and surgery. Machine learning filter models were developed in 18 content categories and were evaluated for discriminatory performance using area under the receiver operating curve (ROC) analysis (AUC). Overall the machine learning filter models identified high quality articles with AUCs ranging from 0.92 to 0.99 across the 18 content categories. Ongoing work addresses direct comparison with Pubmed’s Clinical Query Filters and implementation in a publicly usable resource.

BioPortal: One-stop shopping online for biomedical ontologies and controlled terminologies
Mark A. Musen, Stanford University, Stanford, CA
Monday, November 12, 10:30 am - 12:00 pm
Location: Chicago 9

Nearly all computational tasks in biomedical informatics require, either explicitly or implicitly, the use of standard ontologies and terminologies to provide descriptions of the data on which the tasks operate and the entities that exist in the relevant application area. The rampant proliferation of ontologies and terminologies in recent years has made it difficult for biomedical researchers to know (1) what ontologies are available, (2) what experiences investigators have had using certain ontologies, and (3) whether particular ontologies are suitable for a given purpose. The National Center for Biomedical Ontology, one of the seven National Centers for Biomedical Computing supported by the NIH Roadmap, has developed a Web-based repository of controlled terminologies and ontologies known as BioPortal. Users are accessing BioPortal to browse, to search, and to download biomedical ontolgies and terminologies. BioPortal has become a significant resource for research networks such as BIRN and caBIG. As BioPortal incorporates additional functionality, it will enable its users to rate and comment on ontologies and terminologies. As a result, BioPortal will provide the foundation for a natural experiment in community-based peer-review and evaluation of online knowledge resources. BioPortal may be accessed at http://bioportal.bioontology.org.

Diagnosis and Diagnostic Decision Making: Strategies for Decision Support
Eta S. Berner, University of Alabama at Birmingham, Birmingham, AL
Monday, November 12, 10:30 am - 12:00 pm
Location: Chicago 9

This will describe a special supplement to be published by the American Journal of Medicine. The presentation will describe the analytical review of the literature and a series of commentaries written by experts in diagnostic decision making. The introduction to the supplement is written by Donald A.B. Lindberg.

Microsoft® HealthVault is a new software and services platform aimed at
helping people better manage their health information. HealthVault puts the
consumer in control of their healthcare by creating a private, secure data
storage and sharing platform that will enable seamless data exchange between
hundreds of different health applications and devices. HealthVault Search is
a powerful new vertical health search tool designed to work with the
platform. It intuitively organizes the most relevant online health content,
allows people to refine searches faster and with more accuracy, and
eventually connect them with HealthVault-compatible solutions. Biomedical
informaticians can use the HealthVault SDK to create innovative new consumer
health tools, like personal health records and devices, that allow people to
take meaningful health-related action in the course of daily living.

Bert Van Hoof, Director Consumer Health Solutions at Microsoft, will
describe and demonstrate HealthVault, with commentary by members of the AMIA
Open Source (OS-WG) and People and Organizational Issues (POI-WG) Working
Groups.

Databases are increasingly full of narrative clinical text and structured health records. By correlating phrases in one to facts in the other, we can learn how language is used in health care. We studied the use of temporal assertions in discharge summaries and compared them to events in the electronic health record. For example, whereas "11" weeks, months, or years means eleven units (which might be 10 to 12 in reality), "10" usually means one group of ten with a correspondingly wider variance (say, 5 to 15 in reality). We believe empirical correlation is more broadly applicable to understanding language in health care.

Ethical Challenges in Medical Informatics: Telemedicine and Telehealth
Bonnie Kaplan, Yale University, New Haven, CT, University of Illinois at Chicago, Chicago, IL, and Kaplan Associates, Hamden, CT
Tuesday, November 13, 10:30 am - 12:00 pm
Location: Chicago 8

This presentation highlights ethical, social, and policy concerns in medical informatics. Taking telehealth and telemedicine as an example, the talk is an attempt to broaden discussion beyond privacy, security, confidentiality, and information accuracy. The presentation, based on a forthcoming paper in Cambridge Quarterly of Healthcare Ethics co-authored with Sergio Litewka, MD, of University of Miami, examines informed consent, autonomy, and empowerment in light of issues concerning design, usability, obtrusiveness, and changes in relationships and balances of power among. Societal and policy issues include digital divides; decontextualized or overwhelming volumes of patient information; skilling and deskilling of patients, caregivers, and providers; and shifts in job satisfaction, expertise, and work patterns. My hope is to stimulate more education, discussion, and ethical analysis, not only of telehealth, but also other uses of information and computer technologies that can help improve health care, health, and general well-being.

Failure to Adhere to Informatics: A Case Study of the Oregon RHIO
William Hersh, Oregon Health & Science University, Portland, OR
Tuesday, November 13, 10:30 am - 12:00 pm
Location: Chicago 8

In 2006, a one-year, half-million dollar planning process was undertaken in the Portland, Oregon area to establish a Regional Health Information Organization (RHIO) to facilitate health information exchange across area health care organizations. The process ended in May, 2007 with the rejection by area hospitals and health systems of the proposed plan. Reasons for the failure of the effort included a closed process and a funding model that put the costs largely on local health systems. But an even larger problem was failure to adhere to well-known principles of establishing RHIOs carried out successfully in other communities. This was another example, akin to the increased mortality rate after implementation of computerized physician order entry in the pediatric intensive care unit of Children’s Hospital of Pittsburgh, that a knowledge base for successful implementation of health information technology has been researched and disseminated by the medical informatics community. It demonstrates the larger need to more widely disseminate best practices in HIT implementation and train individuals to apply them.

Translation knowledge into practice: Intellectual inheritance in informatics innovations
Patricia Flatley Brennan, University of Wisconsin-Madison, Madison, WI
Tuesday, November 13, 10:30 am - 12:00 pm
Location: Chicago 8

True, useful informatics innovations rarely emerge fully-formed from a single research project. By tracing the development of the key ideas that formed the basis and products of the National Multiprotocol Ensemble for Self-Scaling Systems for Health project (Kohane, PI), this presentation will identify the intellectual contributions and technical advances afforded by a series of project. Conceptual, methodological, and commercial applications resulted from this line of inquiry and development. The model of “intellectual inheritance” accounts for the contributions of a range of projects to the current state of development. The “intellectual inheritance” model may prove useful to many lines of research that require a way to track the range of efforts that afford the translational results.

Learn about key policy topics. Join us for this late breaking session where we will hear presentations from diverse panelists on a wide variety of public health, applied, clinical, and research informatics topics including proposed and pending legislation, recent regulatory changes, agency strategic priorities and programs relevant to informatics, such as the CTSAs, and research priorities and funding opportunities.

AMIA's mission includes advancing the use of effective clinical information systems. AMIA is aware that Chief Medical Information Officers (CMIOs), Chief Nursing Information Officers (CNIOs) and other “clinical information officers” who work primarily in applied informatics roles to assure that clinical systems are effective, meet the needs of users and promote optimal patient care, are critical to the promulgation of these systems. AMIA wants to be sure that its offerings (e.g., education, networking, etc.), are organized in a way that is of value to this constituency. The goal of the town hall is to a) find out what are the needs of clinical information officers, and b) solicit suggestions for what AMIA might do to meet these needs. The panelists will give their perspectives on the topic. Audience input will be solicited. The results of the deliberation will serve as input to AMIA for future directions.