The National Institutes of Health (NIH) convened a special, 90-minute meeting of the Advisory Committee to the Director (ACD) on May 6 to discuss a concept clearance for establishing a new program, the “Digital Health Equity, Training and Research Consortium.” The consortium would address major challenges and barriers of leveraging artificial intelligence and machine learning (AI/ML) to fuel biomedical innovation.
In opening the meeting, NIH Director Francis Collins, MD, PhD, explained that Congress had specifically appropriated funding to advance research focused on using artificial intelligence (AI) and machine learning (ML) techniques to analyze medical records and other large datasets. The special meeting was called specifically to ensure that a new AI/ML program would be up and funded in the current fiscal year, as required by Congress.
NIH Principal Deputy Director Larry Tabak, DDS, PhD, presented the report of the ACD working group on AI/ML Electronic Medical Records for Research Purposes that supports development of the proposed concept. The working group noted that some especially pressing challenges in the application of AI/ML center on long-standing inequities, including the lack of diverse representation in biomedical studies and datasets, and the prevalence of harmful biases in AI/ML practices and algorithms, which perpetuate and threaten to widen health disparities and inequities. Many under-represented communities lack the financial, infrastructural, and training support to leverage AI in biomedical health and research – despite the fact that those communities have great potential to contribute to data, diverse recruitment and cutting-edge science.
Nina Paltoo, PhD, MPH, assistant director for scientific strategy and innovation in NIH’s National Heart Lung and Blood Institute presented the proposed concept for a multi-year program, with $50 million a year of appropriated funds starting in FY21, aimed at “increasing the participation, representation, and inclusivity of researchers and communities underrepresented in the development of AI/ML models, including electronic health record (EHR) data.” The proposed program aims to develop and support “a researcher and data network of highly diverse institutions that would contribute their own data; leverage AI/ML infrastructure, training, and know-how; and conduct the biomedical research that is most important to them.”
The ACD members unanimously approved the concept, clearing the way for NIH staff to develop a research funding opportunity to be announced this summer, with disbursement of awards in Sept. 2021. A recorded videocast of the May 6 meeting is available. Meanwhile, the ACD is scheduled to meet again on June 10-11, 2021 for discussion of other issues.