One of the key drivers of AI in healthcare has been the shift in policy and regulatory approaches in this sector. There has been significant increases in the number of AI-enabled solutions with FDA approvals or clearance in recent years (Figure 1.)
This is both a driver and sign of the growth of the AI in healthcare sector. FDA has formulated new policies that allow for easier clearance of these technologies. FDA is looking at safety and effectiveness, it will not require patient outcomes . There are people on both sides of the argument on whether the FDA is helping the growth of these technologies by adopting more relaxed policies than the ones used for pharmaceuticals or medical devices, or that it is actually hurting the long-term prospects of the sector.
There are those who believe that since these algorithms change over time and learn from the feedback and with more data, FDA should determine their safety and basic effectiveness, and allow them to be used in the real world setting so the collective experience with these solutions builds up over time. Others believe that The FDA is trying to take a light touch to the regulation of these technologies and although it’s allowing for faster rollout of these technologies, it could lead to significant issues down the road. Examples of this includes technologies that are developed on narrow datasets and underperform when deployed in new medical centers with more diverse patient populations.
These situations can lead to a model that is not adequately trained on enough data and provides false results.
This can lead to health systems applying the brakes on adopting these technologies until the FDA adopts higher standards. This can set the field back several years while these issues are adequately addressed. You can see that this point of view also has some validity and the current FDA approach may actually end up doing some harm to the long-term adoption of these technologies.
FDA’s vision is that, with appropriately tailored total product lifecycle-based regulatory oversight, AI/ML-based Software as a Medical Device (SaMD) will deliver safe and effective software functionality that improves the quality of care that patients receive. On February 7, 2020, FDA announced its marketing authorization, through the De Novo pathway, of the first cardiac ultrasound software that uses artificial intelligence to guide users. This breakthrough device is notable not only for its pioneering intended use but also for the manufacturer’s utilization of a Predetermined Change Control Plan to incorporate future modifications.