Wednesday, June 25, 2008

BioMed Informatics and Engineering Libraries




“Informatics: A Primer”
• Dr. Mary Lou Klem, U. Pittsburgh
• “The study and use of information and communication systems” – Coiera, 2003
• “Analysis and dissemination of data using computers” – NLM, 2005
• “Interdisciplinary study of the design, application, use and impact of IT” – UCI ICS
• Practice: informational and computing sciences (info retrieval, ontology/vocabulary, evaluation/research methods, cognitive/human factors and interfaces, database design, etc.) – Friedman, NIH/NLM, 2004
• Why such a hot topic in health sciences?
• #1 Information explosion – 16 million records in Medline
• 1881: NLM Director predicted geometric progression of medical publication – would require all those not writing it to catalog/index it
• #2 Growth of molecular databases – http://nodalpoint.org
• Librarian role: collection development, education/training, communication – Helms, 2004
• Collections: manage site licensing vs. multiple licenses - 05/06 estimated savings = $200,000
• Education/Training: workshops & consultations in use of molbio software/databases; http://search.hsls.molbio portal utilizing Vivissimo cluster software (tabbed: databases/software, filtered PubMed articles, NCBI genes/proteins, pathways, protocols, recommended articles)
• Communication: hired Head of MolBio Service with PhD in biochemistry, assistants with PhD in biology or neuroscience (and maybe MLIS)
• Search tools:
• Evidence-based practice – clinicians need to consult the “best” evidence when making a patient decision (prognosis, therapy), not what their colleagues or teachers said
• Best = current and most rigorous published studies, varies by type of question
• Few health care professionals have these search skills
• McMaster U. – developed “search queries” or “hedges” – canned searches that were tested for relevancy, accuracy, precision
• Therapy (individual randomized controlled trials) = “clinical trial” in Title/Abstract OR clinical trials in MeSH OR linical trial in Publicatio Type OR random* in Title/Abstract OR “random allocation” in MeSH OR “therapeutic use” in MeSH Subheading
• Prognosis (individual cohort studies or follow-up studies that assess morbidity mortality or survival) = “Incidence in MeSH OR mortality in MeSH OR follow up studies in MeSH OR prognos* in Title/Abstract OR predict* in Title/Abstract OR course* in Title/Abstract
• PubMed: Search by Clinical Study Category – etiology, diagnosis, therapy, prognosis, clinical prediction guides; scope: narrow & specific or broad & sensitive

“Bioinformatics and Engineering Libraries”
• Amy Stout, MIT
• “using computers to solve bio problems, usually on molecular level”
• “new phase in genomics - making bio discoveries not at lab bench but at a computer terminal”
• sequence alignment, evolutionary biology, genome annotation
• Data intensive - mining huge data sets, high through-put data
• Engineering: synthetic biology, sequencing microbes, computational neuroanatomy – circuit diagram of neurons in brain tissue
• Libraries: support researchers and students, data coming under purview of academic libraries, databases like any other (but tools require more specialized knowledge)
• MIT: hired bioinformatics specialist, created website, strengthened collection development, online tutorials, 1.5 hour session on introductory bioinformatics, expert speakers
• Entrez, PubMed, Gene, CoreNucleotide, BLAST (ex: gabra1)
• CoreNucleotide – library catalog-like; each gene has a “bibliographic” record, but also includes actual GATC sequence
• BLAST – find analog to gene in animals (ex: human) other than the one in which you originally found research record (ex: mouse)
• Gene – like review or encyclopedia articles for gene sequences – associated functions (ex: alcohol dependence, history of blackouts, age at first drunkenness, etc.) with links to scholarly literature backing it up

“Biomedical Informatics in Engineering”
• Sheila Young, ASU
• ASU: Dept. of Biomedical Informatics, under School of Computing & Informatics, under School of Engineering
• Bioinformatics (molecular/cellular processes), imaging informatics (tissues/organs), clinical informatics (individuals), public health informatics (populations)
• Bioinformatics: “develop computational tools for the analysis of biomedical data and systems”
• Clinical bioinformatics: “develop novel IT, CS and KM methodologies for disease prevention, treatment, patient care delivery, knowledge access”
o Medical record construction – loss of anecdotal data?
o Medical decision-making
• Imaging informatics: “develop IT and computational tools to manage and analyze biomedical images”
• Public health informatics: “systematic application of information and CS to public health practice, research and learning”
• Engineering – have done similar work for systems biology (modeling and model analysis)
• Computer science, physics, and engineering literature, in addition to PubMed – various thesaurus terms and classification areas
• New interdisciplinary journals
• “Publishing” the data, as well as the narrative

Q: Where does translational science fit in? ISI is releasing a database, new journals, NSF funding.
A: NIH putting huge emphasis on it, funding centers. Pulling researchers from bench to bedside and forcing them to talk to one another.
A: Translational genomics - personal gene reports.

Q: How do the canned searches handle changes in MeSH terms?
A: Assume McMaster is monitoring.

Q: PubMed vs. Medline
A: PubMed contains all Medline records, plus additional or more recent records, from a variety of fields (some may not get indexed) – engineers love how PubMed maps from one database to another.

Q: How handle the $200,000 issue?
A: My understanding is that the library took over the cost from the labs; there are more restrictions on research funding driving this.

2 comments:

Donna Beck said...
This comment has been removed by the author.
Donna Beck said...

As the reporter for the “Bio/Med Informatics and Engineering Libraries” session held at the ASEE 2008 annual conference, I will attempt to highlight some of the efforts that librarians are doing for the biomedical sciences by adding to Amanda Werhane’s great outline. I want to consider how the work pertains to our evolving roles as engineering librarians. I attended this session in an effort to learn about some of the key aspects of bioinformatics that might relate to my liaison role for the Biomedical Engineering and Chemical Engineering departments at Carnegie Mellon University.

Guest speaker Mary Lou Klem from the Health Sciences Library at the University of Pittsburgh introduced the topic of Informatics by pointing out that the field is not just about computers and hardware but also about information science. Informatics is paired with a specialty, for example, “Imaging informatics.” Dr. Klem referred to core skills appropriate to the field of informatics by recognizing ways that librarians should be involved—particularly with their skills in information retrieval and ontology/vocabulary. Informatics looks for ways to use technology to better manage data.

With so many specialized biological databases—over 1,000—librarians help this “very discipline-specific” field by their contributions in collection development, education & training, and communication. Librarians are reminded to look for ways to save money for those individuals in our institutions’ labs and departments by library site licensing of these databases.

The Health Sciences Library System librarians at the University of Pittsburgh provide the typical workshops and one-on-one training on the “MolBio”—molecular biology databases. They also created a portal that serves as a tab-based federated search mechanism, combining their online Bioinformatics Resource Collection with BioMed Central databases HSLS.MolBio
Because of their need for someone with an in-depth knowledge of bioinformatics, the head of their Molecular Biology Information Service has a PhD in Biochemistry.

Dr. Klem then discussed using PubMed for obtaining relevant results useful for the clinician. The user would construct the best possible search by following the steps for “Evidence-Based Practice” in order to obtain the best evidence. A question is posed and then the user needs to define the type of question asked. For instance, the focus of the question might be on the “therapy” needed. For this question, you would be interested in pulling out randomized controlled trials. Conversely, your question might deal with making a “prognosis” in which you would need cohort studies or follow-ups about mortality, morbidity, or survival. Significantly, the search results should not necessarily be limited to what is current, but what is “scientifically rigorous.” As these searches become more complex, PubMed has developed a tool to help. PubMed’s canned searches use search filters by combining controlled vocabulary and key words. These types of searches can be performed by clinicians using the “Clinical Queries” feature. Try a search! Go to: Clinical Queries

Amy Stout from MIT Libraries also emphasized the “data intensive” nature of bioinformatics, pointing out that gene sequences found through work in evolutionary biology can be used when making a comparison to other species. Computers perform the sequence alignment of the DNA. In essence, “databases bleed into tools” that require more specialized knowledge. These tools use complex algorithms to enable, for instance, the ability to design bacteria that can blink. Work at MIT currently includes: Synthetic Biology, Sequencing Micobes, and Neuroanatomy of the Brain.

Along with her colleague, Courtney Crummett, Amy presents workshops on “Bioinformatics for Beginners.” Louisa Rogers, also of MIT, has a guide available at: MIT Bioinformatics
I also found some tutorials on this page:


Amy demonstrated the Entrez search engine that searches across many life science databases:
Entrez cross-database search
She performed a sample search using the name of a particular gene “Gabra1” to show which databases came up with results. Results were found in a number of databases, including a database called: Nucleotide. Here you can find a type of “bibliographic” record for the gene, including its sequence details. Armed with this information, you can go to another database called: BLAST where you can find similarities for the human sequence from the mouse sequence and then also receive an explanation of these matches. Another database, the “Gene” database, is a “collection of review articles for genes.” The records provide a summary about the major functions of the gene.

The main point of Amy’s talk was to show that these sequence discoveries were not found in the laboratory but through use of a computer.

Next, Sheila Young discussed the collaboration that the University of Arizona and Arizona State University have developed to cover the field of Biomedical Informatics. She described Bioinformatics as the “development of computational tools for analysis of biomedical data.” Different types of informatics were detailed, including Clinical Informatics. As librarians, we can relate to the importance in which a medical record is constructed. These algorithms determine the information ultimately retrieved by the clinician. Another example is Imaging Informatics which explores tools to analyze images. Computational tools used in Systems Biology support modeling and modeling analysis in order to learn more about function. Sheila referred to experts in the field whose work is with understanding the functioning of cells by using mathematical models and whose views are from the perspective of engineering systems. Currently “biomedical informatics” is not a MeSH (Medical Subject Heading), and it is not a controlled term in either the Compendex or Inspec databases.

Sheetal Agarwal’s article, “A pervasive computing system for the operating room of the future” that appeared in a 2007 issue of Mobile Networks and Applications represents an example of computer applications being applied to the medical sciences. Sheila notes that “Biology and Medical Computing” is a controlled term in the Compendex and Inspec databases. She also mentioned the IEEE Engineering in Biology and Medicine journal and new journals and databases expected related to Translational Science. The type of research into the genetic basis of human disease involves designing tests associated with risk factors as is being done in the field of Translational Genomics. Other developing fields of study include Infectious Disease Informatics. Ida Sim at Stanford is looking into new ways to manage clinical trials by publishing data as a database rather than a publication. See: Trial Banks The Trial Bank Project

Mary Lou, Amy, and Sheila provided a great introduction to the topic of Informatics. If you are still not convinced on why this subject applies to your own engineering clientele, I encourage you to visit the National Academy of Engineering website:
“Advance health informatics” and “Engineer better medicines” are two of the Society’s “Grand Challenges for Engineering”--problems that engineers have been challenged to help solve as we go forth in the new century.

Submitted by:
Donna Beck
Engineering Librarian
Carnegie Mellon University