|
Abstract: Artificial Intelligence (AI) and healthcare are each other’s top influencer. AI experts believe that healthcare is one of the most important and impactful applications of AI. Conversely, healthcare leaders rate AI as the most important and impactful factor affecting healthcare today. Given this strong reciprocal relationship between AI and healthcare, this relationship must be carefully and properly managed and governed.
AI is contributing to healthcare in many different ways. It is saving costs and time in payments and reimbursement, assisting with appointments and patient communication, automating medical record creation, helping to diagnose and treat diseases, being used to discover new drugs, and has numerous other applications in the health field. But in addition to these beneficial aspects of AI in healthcare, AI presents a number of risks to healthcare, including inaccuracy, unreliability, bias, and privacy concerns.
The U.S. Food and Drug Administration (FDA) has primary regulatory responsibility for health technology, including AI. The agency has been quite proactive in developing new approaches to address some of the novel concerns raised by AI in healthcare while supporting the beneficial applications of AI. But the FDA oversight of AI has two types of gaps. First, there are some areas where the FDA is statutorily prohibited from regulation. Second, even where the FDA does regulate, its regulation is often incomplete, and some additional governance may be needed to supplement the FDA regulation.
These gaps in FDA regulation of healthcare AI can and are being addressed using soft law initiatives. “Soft law” is defined as any measure or program that imposes substantive expectations that are not directly enforceable by government. Soft law comes in a variety of forms, including codes of conduct, principles, private standards, ethical codes, best practices, certification requirements, government non-binding guidance and public-private partnerships. Soft law offers a variety of advantages such as its agility and flexibility to be adopted and modified more quickly than traditional regulation. It is also not limited by agency delegated powers or political jurisdictions, and so has plenary scope that can be applied to any aspect of a problem. Soft law also has its disadvantages though, such as lack of enforceability, public distrust, and limits on participation in developing and implementing the soft law measures.
This article describes and evaluates the role of soft law in governing AI in healthcare. Part I summarizes the current applications of AI in healthcare, as well as some of the problems that have been encountered in using AI in the healthcare sector. Part II analyzes the FDA’s regulatory efforts in governing the use of AI in healthcare, and the governance gaps in the FDA’s regulation. Finally, Part III describes and evaluates current and future soft law initiatives to govern AI in healthcare and discusses possible means to strengthen these soft law programs.
|