Key Dates

NOT-RM-21-021 - Notice of Intent to Publish a Funding Opportunity Announcement for NIH Bridge2AI Integration, Dissemination, and Evaluation (BRIDGE) Center (U54 Clinical Trial Not Allowed)

Office of Strategic Coordination (Common Fund)

This Notice is to alert the community that the NIH plans to publish a Research Opportunity Announcement (ROA) as part of the Bridge to Artificial Intelligence (Bridge2AI) Program to solicit data generation projects to produce flagship datasets for use in biomedical and behavioral science discoveries driven by applications of artificial intelligence and machine learning (AI/ML). The data generation projects will encompass expertise to incorporate into the datasets ethical principles, associated standards and tools, and skills and workforce development to address biomedical and behavioral grand challenges (see illustrative list below). A companion FOA will solicit applications for an Integration, Dissemination and Evaluation (BRIDGE) Center that will integrate, disseminate and evaluate datasets and cross-cutting products across Bridge2AI Data Generation Projects, and will develop best-practices for the use of AI/ML methods in biomedical and behavioral research (see related Notice). It is expected that all Data Generation Projects will work collaboratively with the BRIDGE Center to achieve the goals of the Bridge2AI program.

The Bridge2AI Data Generation Projects (OT2) FOA is expected to be published in June 2021 with a first expected application due date in August 2021. Potential applicants are required to participate in NIH-facilitated teaming activities in the Summer of 2021 to form multi-disciplinary teams to create responsive applications. Earliest award dates will be in March 2022.

This Notice is being provided to allow potential applicants sufficient time to develop responsive applications and to plan participation in the teaming activities. The Bridge2AI website, commonfund.nih.gov/bridge2ai, will provide details and instructions for potential applicants to participate in the teaming activities.

Awards under this funding opportunity will be issued as Other Transaction Agreements (OT2), which are not grants, contracts or cooperative agreements. Other Transactions Agreements will involve active NIH program management. Furthermore, the OT funding mechanism provides NIH with the flexibility to design unique collaborations with private sector entities that may not have experience with commonly used NIH assistance mechanisms such as grants and cooperative agreements. Bridge2AI encourages the involvement of comparatively under-resourced institutions as partners in the data generation projects, as stand-alone applicants or through creative partnership agreements.

The Bridge2AI Program seeks to bridge the biomedical and behavioral research communities with the rapidly growing community of experts developing AI/ML models by producing flagship datasets that adhere to the FAIR principles (Findable, Accessible, Interoperable, Reproducible) and critically integrate ethical considerations in preparing data for computation. The program will use biomedical and behavioral research grand challenges (see illustrative list below) to drive the development of ethics, standards, tools, datasets, and skills and workforce development strategies for linking scientific workflows, protocols, and other information about the data collection process into computable knowledge. Datasets may be linked to existing clinical, environmental and surveillance data as required by the chosen grand challenges. The overall goal of the Bridge2AI Program is to generate flagship datasets and best practices for the collection and preparation of AI/ML-ready data to address biomedical and behavioral research grand challenges. The Bridge2AI program will require multiple scientific domains to come together with data science, data management and analytic experts to unlock the potential of AI/ML for the scientific community.

Planned deliverables of the Bridge2AI program include: 1) New biomedical and behavioral datasets, ethically sourced, trustworthy, well-defined and accessible; 2) Software to standardize data attributes across multiple data sources and across data types (establishing new standards as needed); 3) Automated tools to assist the creation of FAIR and ethically sourced datasets (e.g.: through the intelligent workflows, sensorized instruments, etc.); 4) Resources to disseminate data, ethical principles, tools and best practices; and, 5) Cross-training materials and activities for workforce development that bridges the AI/ML and biomedical/behavioral research communities. Completion of these deliverables will establish a culture for rigorously generating credible, ethical and generalizable data which will enable AI/ML methods to address key biomedical and behavioral grand challenge problems.

The Bridge2AI program is designed to support interdisciplinary Data Generation Projects (this Notice) and a complementary cross-cutting Integration, Dissemination, and Evaluation (BRIDGE) Center (related Notice). Teams funded through these two opportunities will be expected to interact and collaborate regularly to complete Bridge2AI program goals.

The Data Generation Projects (OT2)

Each Data Generation Project will be centered around a biomedical and/or behavioral research grand challenge chosen by each project team to produce relevant multiscale, multi-modal, and multi-stream datasets amenable to AI/ML analyses. The data collected may span the entire spectrum of biomedical and behavioral methods, from atoms to populations, and should be hypothesis-agnostic, ethically sourced, trustworthy, clean, sharable, FAIR, and machine readable. Projects are envisioned as multicomponent undertakings with loci of expertise in related domains, including team science, the relevant data acquisition methods, data and modeling standards, ethical considerations, ancillary tools, and skills and workforce development.

The Bridge2AI program expects project expertise to come from diverse social, cultural, economic, academic and industrial backgrounds and communities. NIH is particularly interested in having Bridge2AI data generation projects that have the potential to characterize the diversity of society and health problems and to have teams that include members from underrepresented researchers, research cohorts and institutions which are less represented as recipients of current NIH funding (see NIH’s Interest in Diversity and information about IDeA states).

This Notice is encouraging potential applicants to begin forming teams around biomedical and/or behavioral grand challenges that promote broad uses for modern AI/ML models. The NIH encourages the community to think across the following illustrative examples of biomedical and/or behavioral grand challenges:

• Digital Twins: Standardizing experimental processes and integrating multiple disparate and heterogeneous sources of data to infer mechanisms to discover emergent properties, while overcoming ethical challenges. A digital twin of ourselves could help us monitor our health histories, predict physical and psychiatric phenotypes, design personalized treatment interventions for multiple health conditions, improve our lifestyles and prevent potential health issues in the future.
• Expanding AI/ML in Clinical Care: Coordinated generation of multimodal data in clinical care and integration of AI/ML models would help to expand AI/ML-assisted diagnosis of individual diseases from a single data source to multiple diseases and bring more advanced in-depth AI/ML prediction models to assist in identifying new biomarkers for early disease detections or risk predictions.
• Functional Genomics: Utilizing emerging experimental and computational genomic approaches, to develop a framework for systematically understanding how coding and non-coding genomic variation connects to disease mechanisms, including an improved understanding of gene pathways and networks. Progress in this area will benefit from integrating genomic data (DNA, transcriptomes, epigenetic data) and other multi-omics data with environmental and clinical data. Development of AI/ML-based predictive models linking functional genomic perturbations to changes in phenotypes is expected to be a central part of this exercise.
• Movement Phenotyping: Creating a suite of validated, reproducible, accessible signatures for overt human behavior (movement) would increase the use of reproducible and objective metrics associated with a host of serious health conditions that include movement as their central indicator. Artificial Intelligence can serve a critical role in developing automated techniques to curate and validate the data to predict diagnostic markers of human movement.
• Precision Public Health: Artificial Intelligence has a strong potential to provide the framework needed to integrate social determinants, which influence health and wellbeing across the life course, reducing structural injustices, social harms and health inequities. A rapid advancement of AI/ML models in health care can both foster more comprehensive technical attention to data and social infrastructures and offer the promise of identifying and applying more effective treatments to populations, which will improve precision public health.
• Salutogenesis: The process by which individuals move from a less healthy to a healthier state, salutogenesis, is likely to involve multiple physiological systems (e.g. cardiovascular, digestive, metabolic, immune), domains (bio-psycho-social) and spatiotemporal scales. The unprecedented grand challenge of understanding salutogenesis will require innovative analytical tools in AI/ML to develop multiscale computational models that explain how physiological and biopsychosocial networks dynamically change over time during the process of human health restoration.

Applications are not being solicited at this time.

Please direct all inquiries to:

Shurjo Sen, Ph.D.
National Human Genome Research Institute (NHGRI)
301 827-7028
[email protected]