Hardware Podcasts

#328 FDA Guidance on Artificial Intelligence (AI) in Medical Devices

Global Medical Device Podcast · 20 Jul 2023 · 50m · Episode 328

What this episode covers

Mike Drues, President of Vascular Sciences, discusses the FDA's April 2023 draft guidance on Predetermined Change Control Plans (PCCP) for Machine Learning Enabled Device Software Functions (ML-DSF). He explains that PCCPs allow AI software to evolve within pre-validated boundaries, replacing the restrictive 'locked algorithm' concept. Drues emphasizes that regulatory logic remains consistent with traditional change management, using examples like surgical cauterizers to illustrate how software can autonomously adjust parameters like temperature while maintaining safety and efficacy.

Most relevant to

early-stage companies

Topics

Mentioned

Companies
Vascular Sciences, Greenlight Guru, FDA
Also
PCCP, ML-DSF, 510K, De Novo

Related episodes from other shows

Other shows on the same ground, closest first.

  • Let's Combinate · 22m

    210 - How AI Is Rewriting GMP: Smarter CAPA, PAT, DOE, and Continuous Yield

    Ben Locwin, a Healthcare Executive and Quality SME, joins host Subhi Saadeh to discuss integrating AI into Good Manufacturing Practices (GMP) within pharmaceutical and medical device manufacturing. They explore current applications like AI-assisted CAPA and deviation management, while highlighting the regulatory challenges of algorithmic opacity in Process Analytical Technology (PAT) and continuous manufacturing. The conversation emphasizes the critical role of Design of Experiments (DOE) in training AI models and using yield calculations as an early warning system for process discrepancies in combination products.

    Quality & yieldValidation & product-market fit
    6 chapters

  • Let's Combinate · 41m

    134 - ISO62304, Software Development Process, SaMD, Closed Loop Systems, Complexities with Cloud and Better Software in MedTech with Richard Koch

    Richard Koch, founder of KochSaMD, explains the complexities of developing cloud-based medical devices, using a closed-loop insulin pump system as a primary case study. He details how ISO 62304 governs software development within the broader ISO 13485 framework, emphasizing the shift from batch-based physical manufacturing to continuous, high-velocity software updates. Koch warns that tech giants like Apple and Google are disrupting the industry by leveraging their digital expertise to acquire or compete with traditional medtech firms, urging leaders to upskill in software development to survive.

    Embedded firmwareConnected products & IoT
    7 chapters

  • Let's Combinate · 36m

    094 - Connected Combination Products, Product Traceability, Regulating New Products, Pre-Determined Changes and AI/ML with Rumi Young

    Rumi Young, Director of Regulatory Policy at Becton Dickinson (BD), discusses the regulatory challenges of smart connected combination products, such as insulin pens and on-body injectors, which integrate sensors, apps, and AI/ML. She explains how unit-level traceability using RFID can optimize manufacturing and clinical trials, while highlighting the current regulatory silos at the FDA that hinder consistent policy for these cross-therapeutic technologies. Young advocates for a single source of truth within the FDA to streamline the use of tools like comparability protocols and predetermined change control for post-market algorithm updates.

    Connected products & IoTAI & machine learning
    8 chapters

  • Let's Combinate · 31m

    206 - Will AI Decide Your Next Drug/Device Approval?

    Dominick Romano, founder of Drainpipe.io, explains how AI can be regulated in pharma and medtech by applying ICH guidelines and process validation to AI systems. He argues that pharmaceuticals are uniquely suited for AI because both deal with 'combinatorial problem sets' involving infinite inputs and outputs, requiring rigorous validation to prevent hallucinations in regulatory dossiers. Romano details how AI can accelerate dossier preparation by reducing millions of working hours in data acquisition and how regulators might use AI agents to speed up review processes, provided the technology is proven safe and accurate by the companies themselves.

    AI & machine learningCertification & compliance
    8 chapters

  • Let's Combinate · 29m

    143 - New FDA Draft Guidance: Essential Drug Design Outputs with Alan Stevens and Rumi Young

    Alan Stevens, Global Head of Complex Devices at AbbVie, and Rumi Young, Director of Regulatory Policy at Novo Nordisk, discuss the FDA’s new draft guidance on Essential Drug Delivery Outputs (EDDOs) for drug delivery devices. They explain the distinction between EDDOs, system-level design outputs, and device-dependent design outputs, emphasizing that this framework applies to both combination products and standalone drug delivery devices like infusion pumps and auto-injectors. The guests highlight the guidance's value in providing concrete examples and clarifying the identification process early in development to ensure consistency across FDA review teams and industry sponsors.

    Medical & health devicesCertification & compliance
    10 chapters

The summary, chapters and topics are generated from the episode’s audio by this directory, not written by Global Medical Device Podcast. The episode itself belongs to its publisher.