In a recent article published by Oxford Academic, authors discussed a number of ways that artificial intelligence (AI) and machine learning (ML) could transform biomedical research and healthcare. This includes enhancing operational efficiency, reducing costs, improving diagnostics, identifying therapeutic targets, and enabling personalized treatment. Despite these opportunities, challenges, such as responsible and ethical implementation, workforce diversity, and equitable access, remain.1
The article makes mention of Monica Bertagnolli, director of the National Institutes of Health, highlighting in a related piece, the need for a multidisciplinary approach involving researchers, clinicians, patients, community organizations, social scientists, equity researchers, and policy experts to optimize AI/ML outcomes.2 Authors also point to President Biden’s recent executive order on AI’s safe development, emphasizing the importance of responsible implementation, considering privacy, security, and civil rights.3
“As Bertagnolli rightfully points out, a multidisciplinary perspective is required to achieve these important goals—one that is inclusive of not only researchers and clinicians but also patients and community organizations, social scientists and equity researchers, and policy and legal experts,” the authors wrote.
ML, as a University of Colorado School of Medicine report notes, can be used to enhance the power of physicians and healthcare professionals, ranging from using closed captioning on a video call with a patient, to something more challenging, such as discovering new personalized medicine treatments for rare diseases.4 ML, the report adds, has evolved at a rapid pace in the last 10 years. CU School of Medicine mentions a 2014 joke about computers taking hours to identify a bird in a photo. Nowadays, a simple phone app can watch a birdfeeder, inform you when one arrives, and identify what type of bird it is.4
Oxford Academic authors acknowledge that there will be further equity challenges in AI/ML implementation, including hurdles in workforce diversity and geographic biases, potential for unintentionally discriminatory algorithms, and possible post-approval application inequities and digital divides.
“To ensure equity, prevent unintended consequences, and maximize AI’s impact and achievements, governance must instead be iterative and dynamic, capable of capturing the broad view of development and evolution of AI across sectors and across all facets of health and medicine,5” the authors wrote. “As described in a recent report from the National Academies and the NAM—Toward Equitable Innovation in Health and Medicine: A Framework—this will require considering the many types of equity in science and technology innovation and how to incorporate them across stages of the innovation life cycle—from conceiving and embarking on an idea, to research and development, to technological evaluation, to access and use of technology, through post-market evaluation and long-term learning. Governance for AI/ML must be able to address the various needs at every stage in the technological life cycle.”
Further improvements to scale and infrastructure are recommended as well. Citing international collaborative efforts, the authors believe that working together will be necessary to achieve scale and avoid costly duplicative efforts. Federal efforts to make this happen, the article cites, include Vice President Kamala Harris’s involvement in the AI Safety Summit. Furthermore, the US Department of State has been heavily involved in the Organization for Economic Cooperation and Development AI Policy Observatory, a platform, the authors explain, that is aimed to shape global public policies for responsible, trustworthy, and beneficial AI. Lastly, according to the article, the US is a member of the Global Partnership on Artificial Intelligence (GPAI), an international and multistakeholder initiative to guide the responsible development and use of AI, grounded in human rights, inclusion, diversity, innovation, and economic growth.
Oxford Academic authors call for a holistic approach to implementation of AI/ML, emphasizing the importance of equity throughout the process. Other vital aspects include making major advances in infrastructure; building out a dynamic, mission-driven governance framework for continuing innovation; and expanding capacity for international collaboration to address the major health challenges of our time.
References
1. Achieving the promise of artificial intelligence in health and medicine: Building a foundation for the future. Oxford Academic. December 19, 2023. https://academic.oup.com/pnasnexus/article/2/12/pgad410/7477225?login=false
2. Bertagnolli, M. Advancing Health Through Artificial Intelligence/Machine Learning: The Critical Importance of Multidisciplinary Collaboration. 2023. PNAS Nexus. https://doi.org/10.1093/pnasnexus/pgad356
3. The White House Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. October 30, 2023. https://www.whitehouse.gov/briefing-room/presidential-actions/2023/10/30/executive-order-on-the-safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence/
4.How Artificial Intelligence is Changing Health Care. University of Colorado Anschutz Medical Campus. October 25, 2023. Accessed December 21, 2023. https://news.cuanschutz.edu/medicine/how-artificial-intelligence-is-changing-health-care
5. Mathews, DJH; Balatbat, CA; Dzau, VJ. Governance of Emerging Technologies in Health and Medicine—Creating a New Framework. 2022. N Engl J Med. 386 (23), 2239-2242. https://www.nejm.org/doi/10.1056/NEJMms2200907?url_ver=Z39.88-2003&rfr_id=ori:rid:crossref.org&rfr_dat=cr_pub%20%200pubmed