The National Library of Medicine (NLM), the world’s largest biomedical library, needs to become the “epicenter for data science” across the National Institutes of Health (NIH), an expert panel concluded yesterday. But it should continue its current programs, ranging from historical exhibits to training for health librarians.
The plan, which encourages new directions for the library but does not recommend that anything be cut, comes as a relief to NLM's diverse stakeholders. Some had been worried that the report would recommend major changes, perhaps even splitting off some of its components and changing its name.
The $328 million NLM has a 179-year history and became part of NIH in 1962. At that point, it was a paper collection with foot-thick walls and 80 kilometers of underground bookshelves. But the library soon added digital resources. In the 1970s, NLM launched MEDLINE, a fee-based database of biomedical abstracts; it later became the freely available PubMed that is used by researchers, medical professionals, and the public worldwide.
NLM also runs GenBank, a repository for genomes and other DNA data, as well as many other databases, such as MedlinePlus, which offers health information for the public, and ClinicalTrials.gov, a registry of clinical trials. Overseeing all of this has been Donald Lindberg, who directed NLM for more than 30 years before he retired in March.
In January, NIH Director Francis Collins asked a working group of his Advisory Committee to the Director (ACD) to come up with a new vision for NLM. Some staff and onlookers worried that NLM might be overhauled into a data science institute. And the group did contemplate a range of ideas, including changing NLM's name, said co-chair Harlan Krumholz, a cardiologist at Yale University, at yesterday’s ACD meeting at NIH in Bethesda, Maryland.
But after learning more about NLM’s programs—which also include standards for electronic health records, toxicology databases, and exhibits on the history of biomedical research—and receiving 650 comments from stakeholders, the panel concluded it's all valuable and that NLM should retain its globally recognized “brand,” Krumholz says.
At the same time, NLM needs to expand its role as a leader in sharing biomedical data in the United States and internationally. For instance, ClinicalTrials.gov could help harmonize standards for such databases, the group said.
And NLM should coordinate data science programs across NIH’s 27 institutes and centers. For example, it should eventually take over NIH's Big Data to Knowledge (BD2K) initiative, a research program aimed at harnessing large data sets that is slated to grow to $100 million a year. BD2K is now part of NIH’s central Common Fund and is run by Philip Bourne, NIH’s first associate director for data science, who joined NIH last year. Bourne’s office will continue to exist even after BD2K moves, Collins says.
The working group didn’t have the time to determine whether NLM needs organizational changes. The next director will need to conduct a thorough review and decide which programs “should be expanded or stopped,” the group’s report says.
Collins accepted the report and has appointed a search committee to find NLM’s new director. As for NLM’s staff, who spent yesterday morning anxiously watching a videocast of the report’s release, they are “very excited about this expansive vision,” said Betsy Humphreys, NLM’s acting director, at the ACD meeting.