Wanted: Big Hairy Audacious Goals

We have a strategic plan. And that plan has three sound and notable goals. So why do we also need a big hairy audacious goal?

The easy answer is that the Blue Ribbon Panel tasked with reviewing NLM’s intramural research program suggested it.

But the real answer pushes us further: To break the limits on our thinking and spark urgency.

Such a bold, risky goal quickens the pulse, sparks excitement, even kindles a bit of fear.

If we are going to achieve that, then we had better get moving.

But what should that be?

The Blue Ribbon Panel offered three ideas (PDF), each building on NLM’s “remarkable track record of research innovation and impact.”

  1. Next-Gen PubMed
    Make PubMed the discoverability engine for the world. Transform it into the single point of access to an array of information types in the life sciences, including data sets, standards, clinical trials, federal health resources, and data science tools and methods. Integrate sophisticated inference capabilities that identify semantic relationships to pull together related content and deliver active learning capabilities and insights, not just hits.
  2. Computational public health
    Create automated tools for disease surveillance and prediction, combining data from disparate sources, including other federal agencies and global partners. Link clinical and epidemiological data to whole genome sequence data for microbial pathogens to rapidly detect, identify, and mitigate the impact of emerging pathogens, pandemics, or malicious attacks.
  3. Artificial Intelligence in medicine
    Build the tools and data management approaches that draw upon large volumes of personal health data to enable automated and precise diagnoses, prognoses, and patient treatment plans.

Whoa. What?!

Any one of these will take years—and a lot of work, skill, coordination, and even luck—to achieve, but then that’s the idea. Big, hairy, audacious goals aren’t meant to be easy. They’re meant to get us reaching beyond what we thought possible.

In the context of NLM’s intramural research program, the Blue Ribbon Panel highlighted several attributes that such goals should possess:

  • Integrate multidimensional data, including temporally dynamic data
  • Impact many fields of biomedicine (including population health)
  • Challenge experts in user interface and user experience
  • Represent difficult multi- and interdisciplinary research challenges
  • Build on unique strengths at NLM and NIH
  • Provide measures of impact and success
  • Require interactions with other agencies
  • Raise profound informatics and data science research questions
  • Represent a substantial engineering challenge for scaling and dissemination

Can we do it?

The Blue Ribbon Panel thinks so. They noted that NLM has achieved tremendous ambitions in the past, including the Unified Medical Language System, the foundational work that enabled CRISPR-Cas, and machine indexing of the world’s biomedical literature.

But is one of their three suggestions the clear and compelling goal that will get us where we want to be?

NLM will be working with stakeholders to examine that question and to identify and lay the groundwork for its future research agenda, but in the meantime let me ask you:

If you wanted to galvanize research across NLM and inspire the larger scientific community, what would you do? Let us know below.


The Blue Ribbon Panel, comprised of nine external experts in biomedical informatics and data science, was asked to look at the following issues:

    • the strengths and weaknesses of NLM’s intramural research program as a whole
    • the quality of it research and training programs
    • the appropriateness of its organizational structure
    • its relationship to other NIH Institutes and Centers
    • its interactions with NLM’s highly regarded and widely used health information services and tools
    • the effectiveness of its review and evaluation processes
    • the suitability of its research facilities.

NLM has already begun reviewing the Blue Ribbon Panel’s recommendations (PDF) and charting a course forward. I’ll keep you apprised of our steps and strategies over the coming year.

Seeking Innovative Methods in Biomedical Informatics and Data Science

Guest post by Dr. Valerie Florance, Director of the NLM Division of Extramural Programs.

It is an exciting time to be a biomedical informatician or data scientist.

First, NLM has committed to transforming the infrastructure of biomedical research and health care. In support of that, NLM’s research grant programs help fund the computational, informatics, and information science aspects of biomedical research to develop and test novel methods with broad application to the research enterprise. NLM supports projects ranging from machine learning to information visualizations, virtual environments, and more.

Second, biomedical data are becoming more readily available. In fact, the production and availability of biomedical data that can be re-used for new research are expanding quickly as NIH works to ensure data generated in NIH-funded research are shared.

Third, Congress has funded several large-scale biomedical research programs at NIH, including the Cancer Moonshot, the HEAL Initiative (Help End Addiction Long-term), and the BRAIN Initiative (Brain Research through Advancing Innovative Neurotechnologies). These represent new opportunities for informaticians and data scientists to bring informatics and data science innovation into the heart of basic biomedical research.

For example, the HEAL Initiative, targeting the opioid epidemic, is studying the biological factors involved in chronic pain. One line of research focuses on discovering and validating biomarkers, biomarker signatures, and endpoints for pain (with a related workshop coming up November 14 and 15, in Washington, DC.). NIH has released multiple funding announcements supporting the HEAL Initiative, with NLM participating in those with computational or informatics elements.

Informatics also cuts across the work of the NIH BRAIN Initiative, which is aimed at revolutionizing our understanding of the human brain. The initiative is focusing its first five years (2016-2020) on developing tools and technologies to advance the field of neuroscience research. The second five-year period will concentrate on applying and refining those tools. A current request for information, open until November 15, 2018, seeks new ideas and directions for the next five-year period.

These two examples of large-scale biomedical research initiatives will benefit tremendously by the extramural community of informatics and data science researchers becoming involved. I encourage your input and participation.

If you do not already subscribe to the weekly update of the NIH Guide to Grants and Contracts, now’s the time to do so, as all new announcements for initiatives such as these will be published there in the coming months. You can also sign up to receive updates specifically about the HEAL Initiative, including webinars and other events.

It is an exciting time to be a biomedical informatician or data scientist, and it’s an even more exciting time to partner with NLM and NIH as we work to accelerate data-driven discovery.

Technology and Data in Mental Health: Applications for Suicide Prevention

Guest post by Elizabeth Chen, PhD, Associate Director of the Center for Biomedical Informatics, Associate Professor of Medical Science, and Associate Professor of Health Services, Policy & Practice at Brown University.

Biomedical informatics as a discipline is broadly concerned with the effective use of data, information, and knowledge to improve human health. Since its origins in the 1950s, we have watched this discipline evolve with advances in health information and communications technology as well as the explosion of electronic health data. During this time, we have also seen the emergence of sub-disciplines reflecting areas of specialization. In fact, a 2015 study uncovered almost 300 different “types” of informatics! Among these was mental health informatics, which first appeared in the title of a 1995 article indexed in PubMed.

Using technology to understand and support mental health dates to the 1950s when specialized television broadcasts delivered mental health training. In the 1960s, computers analyzed data for psychological diagnoses and housed “artificial intelligence” systems that simulated communication with a psychotherapist. More recently, with the rapid adoption of electronic health record (EHR) systems that can collect longitudinal patient information such as diagnoses and medications, we are observing the increased use of EHR technology and data for improving health care, including mental health care.

Mental health remains a global crisis. In the United States alone, mental health conditions affect 1 in 5 adults and children. These conditions are among the factors that contribute to making suicide the 10th leading cause of death overall and 2nd leading cause among 10- to 34-year-olds nationally. With suicide rates having increased by nearly 30% since 1999,  the National Strategy for Suicide Prevention calls for a comprehensive and coordinated approach that includes data-driven strategic planning and evidence-based programs.

There are numerous and wide-ranging applications of mental health informatics and EHRs contributing to these efforts, including the following:

  • Two independent datasets, one including EHR and biobank data from the Vanderbilt University Medical Center, have characterized the role of common genetic variants among those who have attempted suicide. These large-scale genetic analyses support a heritable component to suicide attempts and an incomplete genetic relationship with psychiatric and sleep disorders.
  • At the Parkland Health & Hospital System in Texas, a Universal Suicide Screening Program, initiated in 2012, led to implementing the Columbia-Suicide Severity Rating Scale in the EHR system for adults. The integration of this clinical decision support tool into the clinical workflow demonstrates how technology may be used to improve suicide risk recognition.
  • Researchers across the country are developing models for predicting patients’ future risk of suicidal behavior using “machine learning” techniques, state death certificates, and longitudinal EHR data from a range of health systems, including Partners Healthcare in Massachusetts [PubMed], HealthPartners in Minnesota, Henry Ford Health System in Michigan, and five different Kaiser Permanente locations [PubMed]. Implementing these predictive models as clinical decision support tools in EHR systems has the potential to improve screening, detection, and treatment of suicide risk.
  • In Connecticut, EHR data from the statewide health information exchange and five clinical partners are being used to identify patients at risk of suicide. Claims data from the All-Payer Claims Database and mortality data from the State Department of Public Health will be used to assess the outcomes and impact of the quality improvement efforts.

And these are just a few examples.

Technology and data will continue to play important roles in advancing mental health care. We have already seen the contributions of mental health informatics over the years and those of related areas such as behavioral health informatics and computational psychiatry. There is much more to come in the development of effective and innovative solutions for improving diagnosis, treatment, and prevention of mental health conditions, including those related to suicidal thoughts and behaviors.

headshot of Dr. Elizabeth ChenElizabeth S. Chen, PhD is the founding Associate Director of the Center for Biomedical Informatics, Associate Professor of Medical Science, and Associate Professor of Health Services, Policy & Practice at Brown University. She leads the Clinical Informatics Innovation and Implementation (CI3) Laboratory that is focused on leveraging EHR technology and data to improve healthcare delivery and biomedical discovery. Dr. Chen is an elected fellow of the American College of Medical Informatics and is a member of NLM’s Biomedical Informatics, Library and Data Sciences Review Committee.

 


Dr. Chen will deliver the next NLM Biomedical Informatics & Data Science Lecture on Wednesday, November 14, 2018, at 2:00 pm in the Natcher Conference Center (Building 45), Balcony A. Her talk, “Knowledge Discovery in Clinical and Biomedical Data: Case Studies in Pediatrics and Mental Health,” is free and open to the public. It will also be broadcast live globally and archived via NIH Videocast.

Thank a Medical Librarian

Celebrating National Medical Librarians Month

“Get the word out. Tell the world what we do!”

I received that earnest and heartfelt request from those attending the Medical Library Association’s Midwest Regional Conference in Cleveland earlier this month. And though I thanked the conference attendees for all they do for NLM—helping us connect with our constituents in hospitals, academic institutions, and communities across middle America—I realized there was more I could do to thank and acknowledge all medical librarians, starting with this blog.

I believe that quality information is essential for improved health. It improves clinical decision making and patient care, boosts the quality of biomedical research, supports patients, families, and caregivers, and reduces health care costs.

And who is responsible for organizing and delivering that essential information?

Medical librarians and their partners in the health information profession.

For that, they deserve our thanks, but even more, they should be acknowledged for the myriad ways they improve health care and biomedical research.

Medical librarians…
  • Curate diverse and valuable collections.
    Librarians make deliberate and systematic choices to select the books, journals, data, and other resources needed for research and clinical care.
  • Catalog, index, and make available acquired materials.
    They make the needle you need findable in the collection haystack by adding relevant and appropriate subject headings or keywords to books, journal articles, data sets, images, and other items, which you can then locate by searching freely available databases like PubMed.
  • Manage access rights.
    Medical librarians support copyright and help maintain the intellectual property of authors, publishers, and database creators as they acquire and license resources on behalf of those who need them.
  • Support data discovery.
    Medical librarians identify and create pathways to data repositories that bolster genomic and biomedical informatics research.
  • Find the hard-to-find.
    Librarians know the ins-and-outs of online searching. They’ve trained for it, learning how different databases are organized and how best to extract precise results. Their expertise will save you time and improve outcomes.
  • Help authors publish.
    Librarians can help researchers at every stage of the publishing journey, from writing, revising, and formatting the paper to selecting appropriate and trustworthy outlets for publication.
  • Preserve materials for the future.
    They ensure the collections so painstakingly assembled are safe, secure, and available now and in the years to come, digitizing print materials, monitoring storage conditions, and conserving brittle, crumbling works.

Of course, to thank a medical librarian you have to find one. I suggest starting with NLM’s National Network of Libraries of Medicine (NNLM). At  over 7,000 sites strong, this network provides a point of presence for medical librarianship in almost every county in the US. Many NNLM members are academic institutions, health science libraries, or hospital or clinic libraries, but an increasing number (over 1,700 now) are public libraries taking on new ways to serve their communities.

They’re not alone.

Medical librarians have long ago left the desk behind and stepped into new roles, whether in health care institutions, academic libraries, or private industry. They are leading patient-and-family information services, becoming a part of the knowledge management resources of large health care systems, serving on patient safety and quality control committees, and joining teams of investigators to manage publications, locate critical data sets, gauge research impact, or write grants. From embedded librarian initiatives and innovative outreach programs, medical librarians are deepening the connection with the people they serve, bringing them shoulder-to-shoulder to share knowledge and solve problems.

They’re doing all this because they, too, believe that quality information is essential for improved health, and they know their skills and training put them in the best position to deliver that information.

That’s not only worthy of thanks but of shout-it-from-the-rooftops support. And not just because I say so, but because the data say so.

So, to provide better care, make better decisions, and save money, ask—and then thank—your medical librarian. They’re experts in helping you succeed.

Data in the Scholarly Communications Solar System

Guest post by Kathryn Funk, program manager for NLM’s PubMed Central.

The Library of the Future. What will it look like?  The NLM Strategic Plan envisions it partly as “one of connections between and among literature, data, models, and analytical tools.” In this future, journal articles are no longer lone objects drifting in space, but, rather, each a solar system waiting to be explored. Indeed, we’re already seeing the published literature associated with datasets, clinical trials, protocols, software, earlier versions (including preprints), peer review documents, and so on through consistent identifiers and standardized publishing and archival practices.

To help researchers and the public navigate this new solar system, PubMed Central (PMC), NLM’s full-text archive of journal literature, has been collaborating with publishers and funders for the last year to support efficient ways of linking journal articles with associated data. We’re encouraging authors to cite their open datasets and publishers to archive and make available those data citations in a machine-readable format. Though data citations represent only a small percentage of how PMC articles are linked to data (supplementary material continues to be the predominant method for associating data with articles in the archival record), the growth in data citations in the last year has been promising, nearly doubling the previous year’s total (i.e., 850 articles with data citations in 2017 vs.  approximately 440 in 2016). NLM is also supporting the public access policy requirements of our research funder partners by encouraging authors to deposit datasets as supporting documents via the NIH Manuscript Submission (NIHMS) system.

But solar systems, even the metaphorical kind, are meant to be explored, so we’re also working to expose each journal article solar system in a way that promotes discoverability. We want to make it easier to discover articles in PMC with associated data citations, data availability statements, and supplementary data, through improved record displays and new search facets, leveraging the data-related search filters announced earlier this year.

NLM is also looking beyond datasets to archive and expose articles’ key satellites, including, for example, comments generated during the peer review process. As the effort to expand the openness of peer review gains traction, PMC staff have been collaborating with publishers and Crossref on standardized ways to make readily available those peer review materials.

As with any exploration of new solar systems, it’s our hope that taking these steps will help generate new knowledge, and in so doing drive research that is reproducible, robust, transparent, and reusable. And as we move toward becoming the Library of the Future, how we can best support your research needs in connecting the literature with the rest of the research universe? Please let us know.

With thanks to Jeff Beck for the solar system analogy. 

casual headshot of Kathryn FunkKathryn Funk is the program manager for PubMed Central. She is responsible for PMC policy as well as PMC’s role in supporting the public access policies of numerous funding agencies, including NIH. Katie received her master’s degree in library and information science from The Catholic University of America.