10 Aug Spotlight on the Center for Biomedical Data & Language Processing
Every day research scholars and clinicians around the world generate considerable amounts of biomedical data. The data is produced in multiple formats, including natural languages, medical/biological images, proteomic, and gene expression from various tissues and cell lines. It is impossible for any human being to keep track of all these biomedical records.
Researchers and clinicians track and utilize the information through biomedical data and natural language processing (BioDLP). BioDLP allows documents to be processed from a format where humans understand the information to one where computers are able to re-format it for more effective and efficient end-user utilization. Some applications of BioDLP include: building useful databases of biomedical discoveries from published articles, summarizing research articles, inferring the clinical state of patients from clinical reports, aiding in clinical decision support systems by finding answers for clinical questions from biomedical documents, performing bio-surveillance from clinical reports, and building and enriching ontologies and metadata schemas that can support further document processing. Automatically processing the knowledge present in these documents using BioDLP holds the promise of enhancing the quality of healthcare and also expediting clinical and translational research.
Biomedical data and natural language processing research at the University of Wisconsin- Milwaukee (UWM) continues to be extremely successful. Formal establishment of the Center for Biomedical Data and Language Processing (CBDLP) in May 2011 has allowed for further advancement in the area of biomedical informatics. The CBDLP at UWM has been instrumental in the creation of databases to rapidly analyze biomedical and clinical data.
The TTR spoke with Hong Yu, PhD, founding director of the CBDLP and associate professor in the Department of Health Sciences and Department of Electrical Engineering and Computer Sciences, to get a better understanding of what resources and services the CBDLP offers for CTSI investigators.
What is the mission of the Center for Biomedical Data and Language Processing (CBDLP)?
The mission of the CBDLP is to advance translational biomedical informatics towards personalized medicine. Two long-term projects are collecting genomic data for clinical diagnosis and clinical data mining for patient quality control.
What types of technologies and resources are available to CTSI researchers through your center?
Health care providers often have questions regarding the care of their patients; however, tight schedules leave them with limited time and resources to search for answers. Although some medical knowledge databases are available, such as UpToDate and Thomson Micromedex, studies have found that health care providers often need to consult primary literature for the latest information in patient care. To meet this need, information retrieval systems such as PubMed return lists of retrieved documents in response to user queries, but such searches often yield large amounts of documents. Health care providers usually have limited time to browse retrieved information and studies indicate that physicians are likely to abandon a search if it takes longer than two minutes.
Consequently, the CBDLP created AskHERMES, a computational system which automatically analyzes large volumes of documents pertaining to specific questions and generates short text from them as output. AskHERMES enables physicians to effectively seek information in a clinical setting by applying natural language processing approaches to automatically generate answers in response to ad-hoc questions. AskHERMES promises to provide physicians with the best clinical evidence extracted from primary literature within the time frame demanded in clinical settings.
For example, if a physician inputs the question, “What are the signs, symptoms, and course of Guillain-Barre syndrome?” a substantial search is conducted and completed in only a few seconds. Publications will populate relating to the question, specifying which key word (symptom, course, or signs) it relates to. AskHERMES is an open source online database that is available for clinicians and researchers to use at: http://www.askhermes.org/
How can your Center’s services be utilized to expedite translational research?
The Center is forward thinking when it comes to translational research and hopes to provide quality and rapid up-to-date information for researchers and clinicians. With the rapid development and proliferation in high throughput biological experiments and enormous omic data (including genome, proteome, interactome and localizome), automatic analysis plays an important role in improving the understanding of biological systems. Such improved understanding can be helpful for translational research leading to high-quality patient care. For example, connections between drugs, biological processes, diseases and other adverse effects can be detected by analyzing gene expression data. This could beneficially facilitate several real-life applications of patient care safety, such as detecting drug adverse effects before clinical trial for drug development, and discovering new adverse effects for in-market drugs.
What types of collaboration exist at the CBDLP?
Biomedical data and language processing research is collaborative by nature. To be successful, research in this area requires strong collaborations that include not only computer scientists, informaticians, and linguists, but also biomedical domain experts, like bio-curators, annotators, and practicing clinicians. Being an applied area, its goals and measures of success are often set by biomedical experts. Numerous collaborations exist at CBDLP within the Milwaukee area and beyond. Current collaborators at MCW include faculty from the Human and Molecular Genetics Center.
How can CTSI investigators access your center’s resources and services?
For more information regarding collaboration opportunities or consulting services at the Center please contact Rashmi Prasad, Interim Director at firstname.lastname@example.org.