The future of medical imaging AI depends on the right regulatory environment
March 13, 2020
By Zack Hornberger
The impact of artificial intelligence (AI) on our lives today is remarkable. The adoption of data-driven analytics and the application of machine learning to problems have brought such rapid improvement that value is often understated or overlooked entirely. No industry has benefited more from AI than medical imaging. In just over a decade, medical imaging AI has grown from simple, decision-tree algorithms to complex solutions that hasten the diagnosis of a treatable disease, reduce rescans, and lower radiation dosage.
From software systems that can help identify cardiovascular abnormalities to algorithms that assess the risk of certain neurological diseases such as ALS, the prevalence of these real-world applications will only continue to grow. Equally important, AI by its very nature will improve as more and more data becomes available.
As with any disruptive force in technology, questions inevitably arise. Concerns about patient wellbeing and safety are top of mind. That is why the Food and Drug Administration (FDA) recently held a workshop to discuss the growing role of AI and its impact in medical imaging.
At the Medical Imaging and Technology Alliance (MITA) – the largest trade association representing advanced imaging technology manufacturers – our message is clear. The adoption of AI for medical imaging must support the four pillars of healthcare captured in the “Quadruple Aim”: improving public health, enhancing patient experience, driving cost efficiency, and improving clinician and healthcare staff experience.
To support the Quadruple Aim and take advantage of the promise AI offers, MITA proposes the FDA should ensure that regulatory oversight is consistent across the board. Further, this regulatory oversight should be based on industry Standards and best practices developed by experts with decades of experience in developing and deploying cutting edge innovations. History has shown that balancing necessary regulations for efficacy and safety with frameworks that encourage innovation is an optimal approach when adopting rapidly evolving technologies.
Fortunately, the Agency already has tools at its disposal to facilitate a smooth, transparent pathway for AI technologies. Current regulatory processes, most notably the 510(k) and De Novo classification pathways, are great examples of successful review frameworks that are efficient and flexible without being overly burdensome or stifling for innovators. This is evidenced by the dramatic increase in FDA clearance rates for AI algorithms over the last few years.
Supporting clear review pathways levels the playing field in which competitive innovation is encouraged and restrains regulators from inadvertently choosing winners or losers. Precise and predictable regulatory oversight also ensures that safety and efficacy Standards are fully maintained—a critical component of patient trust.
The promise of AI will be momentous, and the positive contribution to healthcare will extend well beyond medical imaging. The evolving role of this technology should not be a cause for anxiety but rather a reason for immense optimism. MITA will continue to support regulatory strategies that advance Quadruple Aim goals without stifling the innovations that bring the world’s best medical devices to our patients and doctors.
About the author: Zack Hornberger is the Director of Cybersecurity and Informatics at the Medical Imaging and Technology Alliance (MITA)