Skip to main content
    Back to episode
    Episode 10 · January 30, 2025 · 40m listen · 749 words · ~4 min read

    FDA AI Guidance Explained: What It Means for Medical Device Cybersecurity | Ep. 9 - Full Transcript | The Med Device Cyber Podcast

    Read the complete, searchable transcript of Episode 10 of The Med Device Cyber Podcast - expert conversations on medical device cybersecurity, FDA premarket and postmarket guidance, SBOM management, threat modeling, and penetration testing.

    Prefer the listening experience? Open the episode page for the synopsis, key takeaways, topics, and Apple / YouTube listen links.

    Episode summary

    This episode of The Med Device Cyber Podcast delves into the critical implications of Artificial Intelligence (AI) in medical devices, offering essential insights for product security teams, regulatory leads, and engineers. Christian Espinosa and Trevor Slattery explore the history of AI, differentiate it from machine learning, and highlight the distinct risks AI introduces into the medical device landscape. They specifically discuss common attack vectors such as data poisoning, model inversion, model evasion, and performance drift, explaining how these can compromise the integrity, confidentiality, and availability of AI-powered medical devices. The discussion emphasizes the importance of secure development practices, stressing the need to consider cybersecurity from the initial requirements phase through design and postmarket surveillance. The hosts also touch upon the FDA’s guidance for AI in medical devices, including labeling requirements and the challenges of model bias. Key recommendations for manufacturers include rigorous data set vetting, narrowing AI applications, implementing robust guardrails, and continuous postmarket monitoring to ensure consistent and accurate performance. The episode underscores that proactive security measures, implemented "early and often," are paramount for mitigating risks and ensuring the trustworthiness and safety of AI in healthcare.

    Key takeaways from this episode

    • AI and machine learning are related but distinct concepts; AI aims to replicate human intelligence broadly, while machine learning focuses on training computers for specific tasks.
    • Medical device manufacturers should prioritize robust training data vetting and limit AI applications to narrow, well-defined functions to mitigate risks like data poisoning and inaccurate diagnoses.
    • Implementing strong guardrails and input validation is crucial to prevent model inversion and evasion attacks, which could lead to data leaks or incorrect outputs.
    • Continuous postmarket monitoring, including regular performance benchmarking, is essential to detect and address performance drift in AI models, ensuring they remain accurate and effective over time.
    • Adopting a 'security early and often' approach, integrating cybersecurity considerations from the initial design phase, is vital for medical device manufacturers to avoid costly retroactive fixes and ensure product safety.
    • The FDA's guidance on AI in medical devices emphasizes the need for clear labeling and human oversight to address the inherent risks of AI, such as its tendency to 'hallucinate' or produce convincing but incorrect answers.

    Full episode transcript

    Hi, welcome back to The Med Device Cyber Podcast. Today we're going to be talking about an important topic: AI, specifically AI medical devices and some of the risks that AI introduces. We'll also be talking a little bit about the history of AI. We'll go back to 1997 when AI first came out; a lot of people don't realize it. And we'll talk about what manufacturers can do to help secure their devices that have AI in them. We also talk a little bit about some of the attacks on AI and some of the guidance on AI. So, I'm your host, Christian Espinosa, I'm here with Trevor Slattery. How's it going today, Trevor? It's going pretty well. How are you doing today? You know, I did bookkeeping last night. I watched a little bit of our last episode, and I had nightmares about bookkeeping. But I did it last night and did not have nightmares, but I took um, some magnesium before I went to bed, and I think it helps me sleep better. Yep. Do you have that uh, what's that powder called? It's Calm or something like that? You know, I used to take Calm, the gummies. But they stopped selling the gummies; I used to get them at Whole Foods. I don't like the powder, but I like, like the Whole Foods, the gummies. Yeah. Okay, I should check again. I thought they stopped making it with the gummies. Yeah, I guess I've been trying to balance my energy, so like, in the morning, I have like coffee with uh brain octane oil, it's like MCT oil. And then if I have the right kind of nitro cold brew around 1:00 p.m., I have energy the whole day. But then I'm like a little bit wired, so I have to take magnesium to go to sleep. But if I had the wrong kind of cold brew, like there are only certain brands I can drink. If I had the wrong one, I get super irritable the rest of the day. So it's like this, this balancing act, you know. Huh, what's the right brand? It's a Modern Times, San Diego. They only have it at Whole Foods. I tried to order on Amazon, but they don't have it on Amazon. And then every time I go to Whole Foods, they're like out of it, so if I, they have it, I buy like every one of them they have. There you go. Yeah, I've been getting these Yerba Mates at Whole Foods. It's I think it's Peruvian or Brazilian or something like that, but super strong tea, and I can't have any after like 9:00 a.m. or I won't sleep at night. But hey, it works great in the day. Awesome. Well, I guess we're caffeinated and wired, so we can start the podcast talking about AI. So let's, let's kind of start at the beginning and define what AI is. I think there's a lot of ambiguity and confusion about AI, like what it is and how it relates to ML or machine learning. Do you want to explain AI to our listeners, Trevor? Yeah, I think that AI and machine learning are used interchangeably, incorrectly. They are similar and connected, but they're not the same. So AI, Artificial Intelligence, is exactly that. It's something that is trying to replicate human intelligence and human behavior, human process. Machine learning is essentially trying to get a computer to train itself to perform a specific task. So machine learning is effectively a type of AI, but not all AI is machine learning if that makes sense. Yeah, that makes sense. And I know we did a little prep for the podcast, and you mentioned Clippy as one of the first AIs, and we we confirmed that. And I I looked it up, and Clippy was manufactured, or I guess created by Microsoft in 1996 and came out with Office 97, so it's been almost 30 years. Do, do you remember Clippy or do you ever use Clippy? You seem to know a lot about Clippy, but I don't know if you used it before. I caught the tail end of Clippy back in I guess that was Windows Vista, Windows, kind of right when they sunset Clippy. Yeah, so Clippy was that, that paperclip that if you were trying to do something, it would like pop up on your screen, right? And tell you like,