Health sciences educators increasingly rely on multimedia educational materials for their teaching, but creating or finding high-quality multimedia remains a challenge for many. The Health Education Assets Library (HEAL CENTRAL), <http://www.healcentral.org/gioi-thieu>, was formed to facilitate the sharing of existing resources in a freely available digital library, with items organized in a highly searchable database of descriptive information (metadata).

HEAL CENTRAL was established in 2000 as a component of the National Science Digital Library with initial funding from the National Science Foundation. In collaboration with the National Library of Medicine and the Association of American Medical Colleges, the HEAL CENTRAL team has set up partnerships with numerous faculty, medical schools, and other organizations that maintain collections of health sciences resources. HEAL CENTRAL has co-directors at three institutions: Sebastian Uijtdehaage, Ph.D., at the David Geffen School of Medicine at UCLA; Sharon Dennis, M.S., at the Spencer S. Eccles Health Sciences Library at the University of Utah; and Chris Candler, M.D., at the University of Oklahoma School of Medicine.

Currently, the HEAL CENTRAL collection is comprised of a limited set of images, videos and audio files, many of which are premier resources in their fields. Eventually, the library will consist of tens of thousands of high-quality multimedia items, including additional file types, covering a wide scope of health education topics, including patient and consumer health. The goal is a sizable, diversified collection available through a single web-based application that seamlessly accesses a distributed network of related collections. A second grant from the National Science Foundation currently supports this collection development.

Each item in the HEAL CENTRAL digital library is described by a set of metadata, such as title, keywords, usage rights, etc. — twenty-six elements in all, of which eight are required. The HEAL CENTRAL team developed its metadata standard following research into a number of metadata standards developed by national and international standards organizations. The resulting customized metadata schema is based on the Educause IMS, which in turn builds on both the Dublin Core and IEEE Learning Objects Metadata (LOM). Educause IMS is extended to include additional health sciences-related elements, including specimen type, radiograph type, orientation, magnification, disease process, and clinical history.

HEAL CENTRAL resources are indexed with the Medical Subject Headings (MeSH) controlled vocabulary. This hierarchical tree of health sciences-related terms is available within the HEAL application for use by searchers and contributors. Because MeSH is often not specific enough to describe resources in many sub-specialties of medicine, however, the team is now exploring the use of the Unified Medical Language System (UMLS), a meta-thesaurus that maps terminology from many different vocabularies to a central concept, including domain-specific vocabularies such as the Systematized Nomenclature of Medicine (SNOMED).

To provide online access to the multimedia resources in the database, the team has developed a web-based application, available at <http://www.healcentral.org>. A formal requirements analysis process determined which routines and interfaces needed to be created. The finished prototype now allows the user to search, browse, download or upload multimedia files for internal cataloging and approval tasks. Users may search by a simple text string or by combinations of metadata fields. They may also identify materials by browsing a directory organized by MeSH terms. Selected files may be added to a download folder for downloading as a group in a compressed format. Users are invited to contribute multimedia files directly to the HEAL CENTRAL system through a simple web-based wizard. Uploaded items are then added to an internal approval queue; a professional cataloger reviews each item against quality assurance criteria and assigns additional metadata before the item is made available to the public.

For maximal scalability, the application is organized into an n-tiered software architecture. This model of application architecture separates presentation, programming logic, and data into different tiers. The application uses non-proprietary technologies such as Java Server Pages (JSP), Java Beans, Java servlets, and eXtensible Markup Language (XML). The finished prototype is being formally user-tested this spring. Response to HEAL thus far has been excellent, with over 1300 registered users and approximately 60 interested partner organizations. More information on HEAL CENTRAL is available at <http://www.healcentral.org>.

Cre: http://www.dlib.org/dlib/march03/03inbrief.html