What is the Role For AI in Health Care?

With artificial intelligence (AI) gaining visibility as a means of potentially revolutionizing clinical diagnosis and treatment, medical research and health care delivery, the "right tasks" for AI in the health care space were discussed in a viewpoint published Dec. 10 in the Journal of the American Medical Association.

Thomas M. Maddox, MD, FACC; John S. Rumsfeld, MD, FACC; and Philip R.O. Payne, MD, FACC, focus on four key questions: 1) What are the right tasks for AI in health care; 2) What is the right data for AI; 3) What is the right evidence standard for AI; and 4) What are the right approaches of AI integration into clinical care?

In terms of tasks, the authors note that AI is "well suited for those areas of health care where the primary task is identifying clinically useful patterns in large, high-dimensional datasets that cannot be efficiently analyzed by traditional approaches." They write that "AI, used correctly, can serve to augment, rather than replace, human clinical judgement." However, they suggest that "effective AI application for clinical diagnostic and treatment recommendations will require a 'fitness for use' assessment of the data sources and their quality and appropriateness." They also advise that a comprehensive evidence standard, similar to that required for medications and devices, is needed in areas where AI is fueling insights that directly impact clinical care delivery to ensure patient safety.

At the end of the day, Rumsfeld, Maddox and Payne say that AI techniques in their most ideal form "can open up clinical patterns and insights currently beyond human capabilities and free clinicians from some of the cognitive 'load' of integrating the vast and growing corpus of health and health care-related data and knowledge." However, they also caution that "it is essential that AI be recognized for what it can do, and what it can't."

Keywords: Patient Safety, Research, Artificial Intelligence, Information Storage and Retrieval, Therapeutic Human Experimentation, Cognition

< Back to Listings