AMIA Outlines Priorities for NIH Data Science Strategy

Tuesday, April 3, 2018

Nation’s health informatics professionals emphasize the need for FAIR – Findable, Accessible, Interoperable, and Reusable – data practices across all National Institutes of Health grants

BETHESDA, MD – In comments submitted yesterday, the American Medical Informatics Association (AMIA) called on the National Institutes of Health (NIH) to declare that all data generated through its grants align with FAIR data principles. It is not enough that the NIH commit to FAIR data principles, the nation’s experts in health and biomedical informatics contend, the NIH must require that grantees also align to such principles as a condition of funding.

The NIH released its draft Data Science Strategic Plan in early March, articulating several objectives across five goal areas, including: Data Infrastructure; Modernized Data Ecosystem; Data Management, Analytics, and Tools; Workforce Development; and Stewardship and Sustainability. The NIH says this Plan is necessary to capitalize on the opportunities presented by advances in data science, such as machine learning and artificial intelligence, and overcome key challenges, including the growing costs of managing data; siloed data-resources; and misaligned funding streams that create disincentives for improved data sharing.

AMIA strongly supported development of a data science strategic plan, arguing that the NIH must establish itself as a global leader in data science through dedicated stewardship and substantial financial support. Numerous aspects of the Plan were correctly calibrated to achieve the dual goal of capitalizing on data science advances, while addressing long-standing challenges, AMIA said in comments. For example, AMIA noted that the Plan rightly sought to create a “Platform as a Service (PaaS) environment for cloud storage, computational, and related infrastructure services needed to facilitate the deposit, storage, and access to large, high-value NIH data sets.” AMIA also applauded NIH efforts to develop distinct funding strategies for data and software resources, separate from hypothesis-driven research, as well as efforts to develop incentives for grantees who utilize or donate to established/qualified databases and knowledge repositories.

“The NIH is to be commended for seizing a historic opportunity to transform how research is performed and how discovery is translated to care, given our ongoing digital revolution,” said AMIA Board Chair and Regenstrief CEO Peter J. Embi, MD, MS, FACMI. “However, the NIH will need a dedicated assembly of experts to support its ambitious goals. AMIA and the entire informatics community stands ready to help.”

As a foundation for its data science strategy, making all grants align with the principles of Findable, Accessible, Interoperable, and Reusable (FAIR) data practice reveals several additional actions, activities, and policies necessary to achieve such alignment. Some of these actions and activities are captured in the Plan, including the need to provide credit to those grantees who develop useful datasets and software, but several other necessary actions are not.

For example, were the NIH to declare all grantees must adhere to FAIR data principles, then the NIH would need to incentivize such adherence. AMIA recommended that the NIH look to its grants process and update requirements to make Data Sharing Plans “scorable” elements of grant applications subject to the existing policies and include software code and algorithms as required elements within the “rigor and reproducibility” section of grant applications.

“Data sharing has become such an important proximal output of research that we believe the relative value of a proposed project should include consideration of how its data will be shared,” AMIA said in its comments. “These kinds of changes will be force-multipliers for the NIH, as they will encourage broad improvements to how data are collected, shared, and reused.”

To facilitate these changes, the NIH must have qualified experts in biomedical informatics and data science to review applicants’ data sharing plans, AMIA argued, to differentiate between high-quality data sharing plans and low-quality plans. One simple step the NIH should take to identify such expertise is to update the eRA profile templates to capture accomplishments in creating and/or contribute to useful public datasets and software. Additionally, AMIA recommended that the NIH support institutional change within academic promotion and tenure, which continues to rely on publishing journals, by supporting efforts to cite data/software and link such resources to journal articles.

From an organizational capacity standpoint, AMIA strongly urged the NIH to integrate and leverage its existing biomedical informatics programs, grants, workforce training, and education efforts as part of its data science strategy. NIH currently funds and promotes programs within the biomedical informatics arena that have strong overlap with the data sciences, including the NIH Big Data to Knowledge grants, which have supported dozens of university informatics programs, and the recently funded Clinical Data to Health initiative. Additional, AMIA highlighted yearly conferences with direct implications for data science in biomedicine, including AMIA Annual Symposium and AMIA Informatics Summit.

AMIA also noted that the new NIH Chief Data Strategist must have the dedicated authority and resources to compel coordination across the individual Institutes and Centers (ICs). AMIA referenced previous recommendations reiterating that the National Library of Medicine (NLM) play a leading role to foster data science competencies, fund data science tools / services, and otherwise be a central organizational hub for data science across the NIH, which recently finalized a strategic plan of its own.

“The future of clinical research and medicine is awash in data,” said AMIA President and CEO Douglas B. Fridsma, MD, PhD, FACP, FACMI. “If data science is discipline by which knowledge or insights are extracted from data, informatics is the application of such knowledge and insights to biomedical and clinical information systems for the betterment of discovery and patient care. Both disciplines are interrelated, complementary, and necessary to the NIH’s.”

Click here for full comments.

###

AMIA, the leading professional association for informatics professionals, is the center of action for 5,000 informatics professionals from more than 65 countries. As the voice of the nation’s top biomedical and health informatics professionals, AMIA and its members play a leading role in assessing the effect of health innovations on health policy, and advancing the field of informatics. AMIA actively supports five domains in informatics: translational bioinformatics, clinical research informatics, clinical informatics, consumer health informatics, and public health informatics.