Keeping Up with AI-Enabled Medical Device Regulations in 2025

How the FDA is navigating the fast-moving world of AI-enabled medical devices through Predetermined Change Control Plans (PCCPs)

Healthcare is unsurprisingly one of the most tightly regulated fields in the world, and for good reason! These safeguards are in place to protect people, ensuring the care they receive is safe, effective, and of the highest quality. The agency responsible for regulating and approving medical devices in the United States is the Food and Drug Administration (FDA). A medical device cannot be legally marketed or deployed in the U.S. without FDA clearance or approval.

The FDA has regulated medical devices since 1976, but the rise of software as a medical device (SaMD), which includes AI-enabled devices, has created new challenges, as this field is rapidly evolving. By treating SaMD separately from traditional device software, the FDA aims to keep pace with innovation while protecting patient safety (Singh et al., 2025).

Here’s a timeline of AI-enabled medical device approvals in the U.S:

1995: The FDA approved its first AI-assisted device, PAPNET, an AI image processing tool to identify abnormal cells in Pap smears (Singh et al., 2025).

2015: The FDA authorizes 6 AI/ML medical devices (Reuter & Han, 2024).

2024: The FDA authorizes 235 devices (Mesko, 2025)

As of 2025: The FDA’s database now lists 1,247 approved AI/ML-enabled medical devices. (U.S. Food and Drug Administration, n.d.).

In just 10 years, we’ve gone from fewer than 20 AI medical devices to over 1,000.

The FDA first released draft guidance on AI in medical devices a few years ago, with the latest update published on August 18, 2025. One major point is the FDA’s push for AI-device makers to build regulatory submissions around ‘predetermined change control plans’ (PCCPs).

To better understand this complex but crucial topic, I tuned in this week to the Global Medical Device podcast where host Ettiene Nichols spoke with regulatory expert Dr. Mike Drues about the FDA’s latest guidance on Predetermined Change Control Plans (PCCPs) for AI/ML-enabled medical devices.

The Report:

The conversation kicks off with some helpful context, as Dr. Drues dives right into answering the big question: what exactly is artificial intelligence?

He notes that what makes AI truly intelligent is its ability to learn and adapt. Without that, it’s no different from static software we’ve had for decades. Its capacity to adapt is what makes it intelligent.

Dr. Drues then asks: if AI can evolve, how do we set limits to keep it from getting out of control?

This, he explains, is where the Predetermined Change Control Plan (PCCP) comes in as one solution. PCCPs allow companies to make approved updates to AI software without needing to file a new submission. (WCG Clinical, 2025). This helps regulators and companies manage AI-enabled devices by allowing pre-approved updates that balance safety with innovation, which is especially valuable for fast-evolving technologies like those powered by artificial intelligence.

To make this easier to picture, Dr. Drues uses the example of a cauterizer, a surgical tool that cuts and seals tissue to stop bleeding using heat, electrical current, or chemicals. Traditionally, a surgeon would adjust its settings manually, but with AI, the device could sense the procedure in real time and autotune itself within safe, pre-approved limits set by the Predetermined Change Control Plan (PCCP).

As Dr. Drues notes, like physicians learning from experience, AI can adapt on an even greater scale, gaining insights from each patient, applying them to future cases, and potentially sharing knowledge across devices. This represents the future of surgical technology, though for now strict limits on adaptability remain essential to ensure safety and control.

Now how does shifting tasks from a medical professional to an AI-enabled medical device impact regulation?

Dr. Drues explains that the FDA does not regulate the practice of medicine. For example, the FDA has no oversight as to how or when a surgeon manually adjusts a cauterizer. But when the device itself takes over this task through AI software, the FDA needs to step in. Put simply, the responsibility is moving from the surgeon’s hands to the device itself, a step that now falls under regulatory review. To support this shift, new FDA guidance encourages manufacturers to include Predetermined Change Control Plans (PCCPs) in their regulatory submissions.

AI-enabled medical devices are among the fastest-changing areas in healthcare. Regulation has to keep pace, and the Predetermined Change Control Plan offers a way to balance rapid innovation with the need to protect patients as we rapidly charge ahead.

Dr. Drues posed a thought-provoking question with his cauterizer example: if the safety boundaries are the same, does it matter to the patient whether the knob is turned by a surgeon or by software?

For me personally, I’m interested in the FDA-approved AI devices that tap into vast amounts of data to make the most precise decisions possible. What’s fascinating about true artificially intelligent tools is their potential to continuously learn, improve over time, and elevate medicine and healthcare to a new level. With the field evolving so quickly, it’s encouraging to see the FDA updating its regulatory guidance specifically for AI-enabled devices. I’ll be keeping an eye on how these regulations continue to develop as the technology advances. Stay tuned for more!

Here’s the link to the full interview with Dr. Michael Drues on the Global Medical Device Podcast:

Thanks for reading and see you next week!

Resources:

Mesko, B. (2025, July 29). The current state of over 1,250 FDA-approved, AI-based medical devices. Medical Futurist. https://medicalfuturist.com/the-current-state-of-fda-approved-ai-based-medical-devices/

Reuter, E., & Han, J. Y. (2024, October 9). The number of AI medical devices has spiked in the past decade. MedTech Dive. https://www.medtechdive.com/news/fda-ai-medical-devices-growth/728975/

Singh, V., Cheng, S., Kwan, A. C., & Ebinger, J. (2025). United States Food and Drug Administration regulation of clinical software in the era of artificial intelligence and machine learning. Mayo Clinic Proceedings: Digital Health, 3(3), 100231. https://doi.org/10.1016/j.mcpdig.2025.100231

U.S. Food and Drug Administration. (n.d.). Artificial intelligence-enabled medical devices. FDA. https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-enabled-medical-devices

WCG Clinical. (2025, July 11). FDA guidance on AI-enabled devices: Transparency, bias, & lifecycle oversight. https://www.wcgclinical.com/insights/fda-guidance-on-ai-enabled-devices-transparency-bias-lifecycle-oversight/