The President’s Council of Advisors on Science and Technology (PCAST) have released the report Recommendations For Strengthening American Leadership In Industries of the Future (June 2020) https://science.osti.gov/About/PCAST. PCAST advises the President on matters involving science, technology, education, and innovation policy.
Opportunities for NIH https://www.nih.gov to effectively develop and use Artificial Intelligence (AI) is included in the report and reports that much of the prior AI R&D at NIH has focused on text mining, genomic medicine in terms of developing large genomic databases, dbSNP, image processing, and behavioral research.
NIH recognizes the potential of applying machine learning to biomedical data to enable major societal benefits that range from biomedical discoveries and enhancement to provide improvements in health delivery and practice at the community level.
As pointed out in the report, although NIH generates large amounts of data, several challenges exist in linking them effectively with AI data analysis and discovery tools. As stated in the report, NIH does not own much of the data generated by their research funding plus the data is not often subject to privacy and other restrictions. Also, NIH dies not operate or maintain computer facilities, and therefore most of the data is not in a form suited for Machine Learning (ML).
NIH is currently setting up the Artificial Intelligence for Biomedical Excellence (AlBLE) program which is a new NIH-wide Common Fund initiative which will generate new biomedically-relevant data sets amenable to ML analysis at scale. PCAST believes this program has biomedically-relevant data sets amenable to ML analysis at scale, but at the current level of investment will only be able to tackle one problem at a time.
PCAST further believes that it is essential that NIH strengthen partnerships with academia, industry, and other Federal agencies such as DOE, NSF, and NIST in order to meet AI goals in terms of the current COVID-19 pandemic.
The report states, For NIH, there is an urgent need for the vast and highly significant amounts of data currently being generated to be made AI-ready from the start and to manage critical elements of successful collaborations, including common ontologies, data ownership, and access.
To meet the urgent need for NIH to build data management systems to support data governance, provenance, semantics, curation, and assessment of quality, PCAST recommends that every NIH Institute and Center should appoint qualified data science officers to work to achieve this goal.
The report also states that the U.S faces challenges that encompass AI, quantum information science, advanced manufacturing, advanced communications (5G and beyond) and biotechnology. COVID-19 has it made it necessary to advance in technologies needed to support remote learning and to enhance capabilities in medicine and telemedicine.