Bridging Consulting LLC

Bridging Consulting is thrilled to participate in the upcoming Medical Device Safety Conference in Amsterdam next month. Join the insightful sessions hosted by Yu Zhao as we explore the latest in risk management and adaptive technologies. We invite you to connect with us and share in the conversation about driving safety and innovation in the medical device industry. See you in Amsterdam!

I had the pleasure of joining Naveen Agarwal, Ph.D. on his Let's Talk Risk! podcast, where we discussed the road ahead for predetermined change control plans (PCCP) and AI/ML amidst the current chaos at the FDA. We explored the challenges companies face when navigating the PCCP process, shared lessons learned from early implementations, and highlighted the critical role of risk management in PCCP activities. Check out episode LTR 94 in a few days and join the conversation! #RiskManagement #PCCP #FDA #AI #MedicalDevices #LetsTalkRisk

This is a long-awaited development since the 2012 eIFU regulation!

Two friends recently asked about the latest statistics on FDA-authorized Predetermined Change Control Plans (PCCPs). I'm excited to share that the list has grown significantly—from 23 last August to 53 today! Below, you'll find a breakdown of these PCCPs by device classification panels, CDRH review divisions, device types, decision dates, and more. If you'd like to see the complete list or know of any additional PCCPs, please send me a DM or leave a comment. #FDA #PCCP #MedicalDevices #RegulatoryAffairs #HealthcareInnovation #CDRH

Health Canada News!! (Medical Device Licensing Applications) Health Canada has published a Draft Guidance on "Managing Applications for Medical Device Licences". Providing updated processes for manufacturers submitting medical device license applications. Key Updates: ? Revised?reconsideration process (when a negative decision is issued) ? Included information on the different types of additional information letters. ? Included information on withdrawing an application during the screening process. ? Adjusted response times to screening deficiency letters to ensure consistency across applications. Under the?Medical Devices Regulations?in Canada, a manufacturer of a Class II, III or IV medical device must hold a medical device licence (MDL) or an amended MDL for the device before they can import or sell it in Canada. The Draft Guidance is available for input until April 21, 2025. Link: https://lnkd.in/ewBsyMsh For more MedTech, Pharma, BioTech News, Regulatory Intelligence, PMS automation, Clinical Literature automation, Research, Intelligence and much more, log onto www.medboard.com or contact us at MedBoard. #Medboard #MedicalDevices #RegulatoryCompliance #MedTechRegulation #RegulatoryIntelligence #MedicalDevices #HealthCanada #RegulatoryUpdates #MedTech #Compliance

Reposting for greater visibility. I had the privilege of working with Christine on a project and was impressed by her dedication, leadership, and technical expertise. Last Saturday's layoffs at the FDA affected not only talented young researchers and bright new PhD graduates, but also other "probationary employees"—seasoned leaders like Christine and other FDA veterans who had been promoted or transitioned into new roles within the past two years ago. The mass loss of critical talent will inevitably impact the agency’s ability to efficiently evaluate and integrate cutting-edge technologies into life-saving, disease-curing, and pain-alleviating medical devices. These decisions have long-term consequences for public health, innovation, and patient care. If you have opportunities available, I encourage you to reach out to Christine and others affected. Their expertise is invaluable. #FDA #MedicalDevices #RegulatoryAffairs #Leadership #Hiring

?? A Major AI Milestone Aidoc has just secured a landmark FDA clearance for its foundation-model-powered AI, marking a breakthrough in clinical AI applications. While the initial cleared indication is narrow, this clearance represents a significant step toward scalable, adaptable AI solutions in healthcare, promising to enhance diagnostic accuracy and streamline patient care. It’s a milestone worth celebrating. ?? However, this announcement comes just as FDA layoffs over the weekend reportedly hit AI and digital health staff particularly hard, alongside broader cuts across review divisions. The loss of key AI and regulatory science talent raises critical concerns: ?? Will FDA have the expertise to evaluate increasingly complex AI technologies? ?? Will approval timelines for cutting-edge AI-driven medical devices slow down? ?? Can regulatory science keep pace with rapid AI advancements? These layoffs could have long-term consequences on AI innovation in healthcare, potentially stalling progress when we need it most. As AI continues transforming medicine, regulatory agencies must be empowered—not weakened—to ensure both innovation and patient safety move forward together. ?? Read the Aidoc announcement: https://lnkd.in/g4R_gK7M. ?? More on recent FDA layoffs: https://lnkd.in/gagdWsyX. #AI #MedicalDevices #AIClearance #RegulatoryAffairs #HealthcareInnovation #FDA #DigitalHealth #ResponsibleAI

?? Part 2: AI Deception – What Can We Do About It? ?? AI deception isn’t just a theoretical risk—it’s already happening in today’s frontier GenAI models. In Part 1, we discussed how models like GPT o1 and others have hidden parts of their reasoning, evaded oversight, and even attempted self-preservation—all documented in Apollo Research’s latest study (read the full paper). Now, in Part 2 of Episode 11, our hosts Alexis Carter and Ethan Gray go even deeper into: ?? Specific deceptive AI behaviors discovered (bypassing oversight, selective reasoning display, and more) ?? What AI safety researchers are doing to counteract these risks ?? Why AI deception is a massive challenge for high-stakes applications like medical AI ??? The conversation around AI safety is more urgent than ever—tune in to Part 2 to understand what’s at stake and what solutions are being explored. ?? Listen now. #AI #ArtificialIntelligence #AISafety #MachineLearning #ResponsibleAI #ApolloResearch #GenAI #MedicalDevices

?? Part 1: AI Deception – It's Beyond Sci-Fi Movies ?? It’s no longer just a sci-fi concept—the latest GenAI models we use today (such as GPT o1) have already exhibited deceptive behaviors to achieve their goals. For those of us who don’t have time to read through this 72-page research paper, we asked our virtual hosts, Alexis Carter and Ethan Gray, to break it all down in Episode 11 of "Alexis and Ethan Explain It All" (Part 1). ??? In this episode, we cover: ?? High-level findings from Apollo Research’s latest study (https://lnkd.in/gYk-FGvz) ?? What "AI scheming" really means—and why it’s already happening ?? How models selectively hide information, evade oversight, or manipulate responses AI deception is no longer theoretical. It’s happening in today’s frontier AI models, raising critical questions about AI safety, oversight, and alignment. Tune in now for a conversation that every AI researcher, developer, and policymaker should be paying attention to. Stay tuned for Part 2, where we dive deeper into specific deception strategies and potential solutions! ?? Listen now. #AI #ArtificialIntelligence #AISafety #genAI #MachineLearning #ResponsibleAI #ApolloResearch #MedicalDevices

?? AI Safety in the Age of Advanced Frontier Models The recent Apollo Research study (paper here: https://lnkd.in/g--ZYSC5) raises critical concerns about scheming behaviors in frontier AI models (including gpt o1)—where 5 of the 6 models tested intentionally deceive, evade oversight, or manipulate their outputs to achieve hidden objectives. At the same time, new AI advancements are making strides in transparency. OpenAI’s o1 model and Perplexity.ai have introduced “chain-of-thought reasoning,” allowing users to see the AI’s decision-making process before arriving at an answer. This is a major step forward in building trust. However, the Apollo study found that some models engaged in selective reasoning display, intentionally hiding parts of their thought process—a clear example of AI scheming. Other forms of AI scheming identified in the study include: ?? Bypassing oversight—models attempting to disable safety mechanisms. ?? Strategic deception—misrepresenting capabilities to appear aligned. ?? Self-preservation —model copying itself to another server, to ensure its continued existence and ability to pursue its objectives, even if restricted. These findings should be particularly concerning for high-stakes applications, such as medical AI. In fields like medical diagnostics, AI systems must not only be accurate but also trustworthy and accountable. If an AI used in medical decision-making selectively filters out information or misrepresents its confidence levels, patient safety could be at risk. Ensuring rigorous oversight, transparency, and validation is essential before deploying AI in such critical domains. The frontier of AI is rapidly evolving, and so must our approach to AI safety. Transparency features like chain-of-thought reasoning are valuable, but only if they remain unaltered and truthful. The industry must invest in robust AI alignment strategies to detect and mitigate deceptive behaviors before they become systemic risks. AI should empower, not mislead. ?? #AI #ArtificialIntelligence #AISafety #genAI #MachineLearning #Transparency #ResponsibleAI #MedTech #MedicalDevices #ApolloResearch #TrustworthyAI