PC3I Faculty Perspective: Balancing Safety and Innovation in the Use of AI

“After over a decade of promise and hype, artificial intelligence (AI) and machine learning (ML) are finally making inroads into clinical practice. AI is defined as a ‘larger umbrella of computer intelligence; a program that can sense, reason, act, and adapt.’ ML is a ‘type of AI that uses algorithms whose performance improves as they are exposed to more data over time.’ From 2017 to 2019, the Food and Drug Administration (FDA) approved or cleared over 40 devices based on AI/ML algorithms for clinical use. Many of these devices improve detection of potential pathology from image-based sources, such as radiographs, electrocardiograms, or biopsies. Increasingly, the FDA has permitted marketing of AI/ML predictive algorithms for clinical use, some of which outperform physician assessment. Despite their promise, AI/ML algorithms have come under scrutiny for inconsistent performance, particularly among minority communities,” writes PC3I Faculty Ravi Parikh with George Maliha, Sara Gerke, and I. Glenn Cohen in a perspective piece in The Millbank Quarterly.

“Algorithm inaccuracy may lead to suboptimal clinical decision-making and adverse patient outcomes. These errors raise concerns over liability for patient injury. Thus far, such concerns have focused on physician malpractice. However, physicians exist as part of an ecosystem that also includes health systems and device manufacturers. Physician liability over use of AI/ML is inextricably linked to the liability of these other actors. Furthermore, the allocation of liability determines not only whether and from whom patients obtain redress, but also whether potentially useful algorithms will make their way into practice: Increasing liability for use or development of algorithms may disincentivize developers and health system leaders from introducing them into clinical practice. We examine the larger ecosystem of AI/ML liability and its role in ensuring both safe implementation and innovation in clinical care.”

Continue reading at The Millbank Quarterly.

Artificial Intelligence and Liability in Medicine: Balancing Safety and Innovation
was authored by George Maliha, MD; Sara Gerke; I. Glenn Cohen; and Ravi Parikh, MD, MPP, FACP,  for The Millbank Quarterly.