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.”
- 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.
- 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.
- 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.
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.
6 thoughts on “Wanted: Big Hairy Audacious Goals”
I’m not familiar with the Blue Ribbon Panel, but I can see from their suggestions that they are from the biomedical informatics field. To attain goals such as computational public health and AI in medicine, NLM needs a different work force. And, in building a work force to attain these goals, I question what the role of our health sciences librarians will be in these projects. What opportunities are there for my profession in the direction that NLM is taking?
To inspire big ideas, I would encourage the community to read outside their traditional fields of study and to wonder, always, what if…
I wonder, too, about the use of terms “artificial intelligence” and “augmented intelligence” (as used by the American Medical Association in their first policy recommendation on AI). Will AI in medicine help professionals work more efficiently? Can it help practitioners focus more on relational aspects of care?
Lastly, I would engage with legal and ethical experts to discuss potential concerns and take opportunities to step back from the work and think about real-world impact.
Response by Vojtech Huser
The following project would be a nice goal
Title: Healthcare consumer view of BigData in Healthcare
Description: Learning Health System accumulated de-identified data on healthcare events in order to learn over time and improve. A patient (health consumer) or patient relative may feel left out and may not see the benefit of this BigData for him/her. This initiative would try to show to the consumers (by creating consumer friendly data views) how this data can help them. E.g., A patient was just prescribed drug X from his internal medicine doctor – he would go to a web based system and see as much as possible we can offer to the consumer about drug X from BigData in healthcare. So under pubmed search for drug X, [s]he would see the data view about drug X from Medicaid/Medicare/Other data. Example for UK is https://ebmdatalab.net/open-prescribing But not just drugs are target for consumers. E.g., diagnoses, lab tests, procedures