Researchers at the University of Virginia School of Medicine have created a new online search engine-www.relemed.com-that provides medical professionals, researchers and the general public with a more efficient and targeted way to search PubMed for the latest, most relevant medical literature to answer medical queries.
ReleMed-short for Relevant Medicine-is not a general health site. It doesn’t provide answers or suggest guidelines for specific medical problems. Rather, based on search terms, ReleMed retrieves the most relevant recent references published about a problem or a combination of conditions, versus any article in which the search terms appear.
“Most of the search engines that examine the 16 million articles currently indexed by The National Library of Medicine give you the most recently published articles first, but they don’t look for relevance of the articles to your query,” said Dr. Mir Siadaty, ReleMed’s developer. “For example, say you want to find the most relevant articles for the relationship between Hepatitis C and Arthritis. If you search using PubMed you will receive hundreds of articles that randomly contain the terms ‘Hepatitis C’ and ‘Arthritis’ but without, necessarily, indicating the relationship between them. Even if the PubMed search happens to return articles in which the relationship between Hepatitis C and Arthritis is discussed, those articles can be hard to find. You may have to read through dozens, sometimes hundreds, of abstracts to locate them.”
“We designed ReleMed so that the first articles returned in response to a query have close relationships between the search terms. This means you can quickly find not only the most recent articles, like PubMed does, but also-and more importantly-the most relevant articles.”
ReleMed works by assigning each of the articles a relevance or priority score. An article assigned a ranking of “1”-the highest possible-has clear relationships among the submitted search terms in all the critical parts of the article: the title, the abstract and in the key indexing or MeSH terms.
“ReleMed still lets the user decide which articles they want to examine in more detail,” said Dr. William Knaus, professor and chairman of UVa’s Department of Public Health Sciences and Siadaty’s collaborator, “but it makes that process so much quicker and more efficient by ranking and then color-coding the results.
“With the explosion in the number of biomedical articles being published (more than 1,700 new articles daily), ReleMed allows you to cull through them efficiently. This great new tool retrieves and then organizes articles on medicine and biology so that the user spends less time reading through articles that are not directly responsive to their needs.”
In an article recently released as a pre-print from JAMIA, the Journal of the American Medical Informatics Association, researchers from Houston and Portland examined more than two million PubMed searches and concluded that the majority of PubMed users seek information on the search terms and their relationship to one another_which is what ReleMed provides_rather than a slew of articles in which the search terms randomly appear. As reported in the article (http://www.jamia.org/cgi/content/abstract/M2191v1), they also found that the majority of PubMed queries entail multiple search terms and that a relevance metric-of the type ReleMed provides-is needed to help users find articles faster.
Even expert searchers such as librarians have been frustrated with the inability to sort MEDLINE search results by relevance. Two medical librarians at the University of Virginia’s Claude Moore Health Sciences Library have been using ReleMed over the past year. Karen Knight, Medical Education Librarian, notes, “Without a relevance filter, the searcher must scan all the citations and manually sort the relevant from the less relevant hits. ReleMed allows you to view your search results by relevance.” “By presenting the best results first, it makes scanning the results easier for both me as a search professional, and for my clients, from novice to expert.” adds Andrea Horne, assistant director for Information Services.
ReleMed is easy to use. Unlike most current MEDLINE search engines, it requires no special training or multiple steps. It accepts common medical terminology. Simply enter diagnoses or medical problems in the search box, and ReleMed returns a list of references with those search terms highlighted in red. A green box appears next to the title of the article; the darker the box, the higher the relevance. Hover the mouse over the box, and the exact relevance level appears, along with information on why ReleMed has assigned it.
With ReleMed, clinicians and researchers with tightly defined search needs can quickly search for articles that address their specific interest.
According to Siadaty, ReleMed’s improved search function is the first step in making it easier for users to retrieve the right information the first time. “We are constantly improving ReleMed by adding new capabilities, and we welcome feedback from both the medical profession and the public.”
ReleMed users can provide feedback from the site: www.relemed.com. Technical information can be found at http://www.biomedcentral.com/1472-6947/7/1 where the journal article describing ReleMed’s construction has become one of the most highly accessed articles on the site.
