Cosm

AI/ML model performance can shift over time for various reasons. In fact, AAMI 34971 calls this out as a risk in AI/ML models for medical devices. So how do we mitigate it? Post-market monitoring can help identify drift. As the integration of machine learning in medical devices expands, AI/ML developers face new challenges, with data drift being a critical concern. Data drift refers to changes in the data distribution between the training phase and real-world application, significantly impacting model performance. In our lastest blog post, we explore: ?? What is Data Drift? ?? Types of Data Drift ?? Implications of Data Drift in Medical ML ?? Mitigation Strategies (Pre-deployment and Post-deployment) Learn how to tackle data drift and ensure your ML models maintain high performance and trustworthiness in clinical settings. Don’t miss out on this essential read for AI/ML developers working in the medical field! ?? See the post here: https://lnkd.in/grPSrCsv #MachineLearning #AI #MedicalImaging #DataDrift #AIML #HealthcareInnovation #MedicalDevices #fda #cdrh #regulatory #digitalhealth #artificialintelligence #machinelearning #samd #aidrift Image Source: FDA website

?? Bridging the AI Literacy Gap: The AI Pact’s Living Repository ???? As AI adoption accelerates, ensuring AI literacy across industries is more critical than ever. The EU AI Act (Article 4) mandates providers and deployers of AI systems to equip their teams with sufficient AI literacy—helping businesses navigate compliance while fostering responsible AI use. To support this, the EU AI Office has launched a Living Repository showcasing real-world AI literacy practices from organizations across sectors. This repository is a growing resource for businesses to exchange knowledge and strengthen AI competency. ?? What’s inside? ? Fully implemented AI literacy programs from leaders like Booking.com, EnBW, and Telefónica ? Partially rolled-out initiatives from firms in healthcare, finance, and public administration ? Planned AI literacy strategies for organizations preparing for the AI Act’s impact A few standout examples: ?? One company designed a custom AI training series for its legal and public affairs teams, ensuring they grasp AI’s regulatory impact. ??Another company employs game-based AI training to engage employees at all technical levels. ?? Another company integrates Responsible AI Literacy into its corporate culture, ensuring compliance and ethical AI use across its business. ?? Why does this matter? AI literacy isn’t just about understanding models and algorithms—it’s about empowering employees, mitigating risks, and ensuring ethical AI deployment. The Living Repository provides a blueprint for businesses looking to enhance their AI literacy programs. ?? Explore the repository here: https://lnkd.in/exu_BxvD ?? Have you implemented an AI literacy program in your organization? Share your thoughts below! ?? #AIAct #AILiteracy #ArtificialIntelligence #Regulation #Compliance #DigitalTransformation #AIForGood

?? AI in Radiology: Preparing for the EU AI Act ?? A European Society of Radiology (ESR) AI working group has just released an important recommendations on effective implementation of the EU AI Act in medical imaging. ?? Key Takeaways from the ESR Report: ? AI in radiology is high-risk by default – requiring robust oversight, data governance, and post-market monitoring. ? AI literacy is critical – radiologists, healthcare professionals, and even patients need better education on AI’s benefits, limitations, and ethical considerations. ? Human oversight remains essential – automation bias and deskilling risks must be addressed through continuous training. ? Quality management alignment – the AI Act must harmonize with EU MDR to simplify compliance for AI-powered medical devices. ? Regulatory sandboxes – while innovation should be encouraged, AI in medical imaging should undergo rigorous, large-scale clinical studies before approval. ? Transparency & accountability – AI model cards and standardized reporting are needed to ensure that AI tools are explainable and clinically reliable. ?? What’s next? The ESR calls for close collaboration between policymakers, regulators, and the medical imaging community to ensure AI adoption enhances patient care without compromising safety or quality. ?? Read the full ESR statement here: https://lnkd.in/gcrX63cv #AI #Radiology #MedicalImaging #MedicalDevices #Regulation #EUAIAct #HealthcareInnovation #ArtificialIntelligence #ESR #SaMD #EUMDR

?? New Opportunity for Medical Device Manufacturers: EMA’s Scientific Advice Program ?? Big news for medical device manufacturers navigating the EU regulatory landscape! The European Medicines Agency (EMA) has just established a regular procedure for obtaining scientific advice on high-risk medical devices (Class IIb and III). ???? This initiative allows manufacturers to engage with EMA and expert panels early in their development process—a game-changer for those seeking guidance on clinical evidence, study design, and regulatory expectations. Why This Matters ?? Clarity on Clinical Evidence Requirements – Avoid regulatory roadblocks by aligning with EMA’s expectations. ?? Early Engagement = Fewer Surprises – Address potential concerns before costly clinical trials. ?? Faster Market Access – Strengthen your submission with well-supported clinical data. Who Should Pay Attention? ?? Companies developing high-risk devices requiring expert panel consultation under the MDR. ?? Manufacturers seeking EMA’s guidance before submitting clinical investigation data. ?? Regulatory professionals and strategists aiming for smoother authorization pathways in the EU. ?? Read more: https://lnkd.in/gyzuafWx This is a crucial step toward greater regulatory predictability and better patient outcomes. #MedicalDevices #Regulatory #RegulatoryAffairs #EMA #EURegulations #ClinicalEvidence #notifiedbody #MedicalTechnology #MedTech #MDR #cosmhq

QMS Audits for MedTech & AI Companies Compliance is the foundation of success in the medical device industry, and staying ahead of evolving regulatory expectations is more critical than ever. At Cosm, we’re excited to announce that we are now offering Quality Management System (QMS) audits to help medical device manufacturers and AI developers ensure compliance with: ? 21 CFR 820 - FDA’s QMS requirements for medical devices ? ISO 13485 - Global standard for medical device quality management ? MDSAP - Medical Device Single Audit Program, streamlining compliance across multiple jurisdictions ? ISO 42001 The new AI Management System Standard, essential for developers of AI-driven healthcare solutions Why is this important? A well-structured QMS is not just about regulatory approval—it’s about patient safety, product reliability, and market success. With increasing scrutiny on AI/ML-driven medical technologies, adherence to ISO 42001 is becoming a key differentiator in regulatory submissions and market access. Compliance to ISO 42001 will also help with the EU AI Act How can we help? Our expert auditors will assess your QMS, identify gaps, and provide actionable insights to help you achieve and maintain compliance—ensuring smoother regulatory approvals and reduced business risks. If you’re interested in partnering with us for an internal audit or regulatory gap assessment, contact us at [email protected]. Let’s ensure your quality, compliance, and innovation are built to last! #QualityManagement #MedicalDevices #RegulatoryCompliance #ISO13485 #21CFR820 #MDSAP #ISO42001 #AIinHealthcare #MedicalAI #QMS #FDA #MedTech #HealthcareInnovation #Audit #RiskManagement #AIRegulation #CosmHQ #ArtificialIntelligence #AIA #AIAct

Expanding Our Services: QMS Audits for MedTech & AI Companies Compliance is the foundation of success in the medical device industry, and staying ahead of evolving regulatory expectations is more critical than ever. At Cosm, we’re excited to announce that we are now offering Quality Management System (QMS) audits to help medical device manufacturers and AI developers ensure compliance with: ? 21 CFR 820 - FDA’s QMS requirements for medical devices ? ISO 13485 - Global standard for medical device quality management ? MDSAP - Medical Device Single Audit Program, streamlining compliance across multiple jurisdictions ? ISO 42001 The new AI Management System Standard, essential for developers of AI-driven healthcare solutions Why is this important? A well-structured QMS is not just about regulatory approval—it’s about patient safety, product reliability, and market success. With increasing scrutiny on AI/ML-driven medical technologies, adherence to ISO 42001 is becoming a key differentiator in regulatory submissions and market access. Compliance to ISO 42001 will also help with the EU AI Act How can we help? Our expert auditors will assess your QMS, identify gaps, and provide actionable insights to help you achieve and maintain compliance—ensuring smoother regulatory approvals and reduced business risks. If you’re interested in partnering with us for an internal audit or regulatory gap assessment, contact us at [email protected]. Let’s ensure your quality, compliance, and innovation are built to last! #QualityManagement #MedicalDevices #RegulatoryCompliance #ISO13485 #21CFR820 #MDSAP #ISO42001 #AIinHealthcare #MedicalAI #QMS #FDA #MedTech #HealthcareInnovation #Audit #RiskManagement #AIRegulation #CosmHQ #ArtificialIntelligence #AIA #AIAct

?? New MHRA Guidance on Digital Mental Health Technologies (DMHTs)?? The UK MHRA has just released long-awaited guidance for digital mental health technologies, a critical step toward ensuring that apps, AI-driven platforms, and other digital tools meet regulatory standards for safety and effectiveness. ?? Key Takeaways: ? Is your DMHT a medical device? It depends if it has a medical purpose and sufficient functionality. Sufficient functionality has two categories: low functionality and high functionality where only the latter would likely be a SaMD. ? Classification matters. The risk level of your technology (Class I, IIa, IIb, or III) will determine the regulatory pathway, with AI-driven or clinical decision-support tools facing stricter requirements. ? Compliance is non-negotiable. UKCA or CE certification, post-market surveillance, and cybersecurity measures are essential to ensuring patient safety and regulatory authorization. ?? What’s next? If you’re developing a DMHT, now is the time to assess compliance, refine regulatory strategies, and ensure your technology aligns with UK medical device regulations. Check out our blog post more a more in-depth look - https://lnkd.in/gR2bTnBk ?? What are your thoughts on this new guidance? Will it help bring safer, more effective digital mental health solutions to the UK market? Let’s discuss in the comments! ?? #DigitalHealth #MentalHealthTech #MedicalDevices #AIinHealthcare #RegulatoryAffairs #HealthTech #MHRA #SaMD #Cybersecurity #UK

?? Breaking News for the IVD & Healthcare IT Community! ?? The long-awaited CLSI AUTO11 Edition 3 has officially been released! ?? This updated standard is a game-changer for IVD manufacturers, healthcare delivery organizations (HDOs), and software developers working to enhance the security of in vitro diagnostic (IVD) instruments and software systems. ?? Why is this important? As healthcare IT continues to advance, cybersecurity risks are evolving just as fast. With the growing connectivity of IVD devices—many of which rely on networked environments, cloud integration, and remote support—it is critical to have robust technical and operational IT security standards in place. CLSI AUTO11 provides industry guidance to ensure data security, system integrity, and compliance with best practices. ?? What does CLSI AUTO11 cover? ? IT security requirements for IVD systems (e.g., devices, analytical instruments, data management systems) ? Implementation guidance for manufacturers & HDOs ? Best practices for mitigating cybersecurity risks ? Considerations for legacy systems vs. modern networked environments ? Alignment with international security standards (ISO, IEEE, etc.) ?? What’s next? While CLSI AUTO11 Edition 3 is not yet FDA-recognized, it’s likely just a matter of time, given that Edition 2 was officially recognized. IVD manufacturers and HDOs should start familiarizing themselves with the latest updates to stay ahead of evolving cybersecurity challenges. ?? Although this is specific to IVDs, the rest of the industry can probably take now and apply similar practices. Reach out if you have any questions - [email protected] #IVD #Cybersecurity #HealthcareIT #MedicalDevices #DataSecurity #CLSI #RegulatoryCompliance #IVDManufacturing #DigitalHealth #FDA #invitrodiagnositcs

?? Streamline Your Software V&V Efforts for Medical Devices ?? In today’s fast-paced digital world, software is the backbone of medical devices—and ensuring its reliability is critical to patient safety. That’s where Software Verification & Validation (V&V) comes in. ?? But where do you start? Creating a robust V&V plan, protocol, and report from scratch can be overwhelming. Without a structured approach, teams risk inefficiencies, compliance issues, and delays in regulatory authorization. ? We’ve got you covered! To help medical device developers navigate this process efficiently, we’ve created ready-to-use V&V templates aligned with FDA, EU MDR, and IEC 62304 standards. Our blog dives deep into the essential steps for designing a strong V&V strategy—covering risk-based testing, requirement traceability, and test case development. ?? Read the blog & get the templates here: https://lnkd.in/ea8BuuV4 ?? Are you facing challenges in your SaMD V&V process? Let’s discuss [email protected] #SoftwareVandV #FDA #EUMDR #MedicalDevices #SaMD #SoftwareTesting #RegulatoryCompliance #QualityAssurance #MedicalDeviceDevelopment #FDACompliance #MDR #IEC62304 #MedTech #RiskManagement

Navigating IMDRF’s Final Guidances on Risk and AI. See our blog post for more but here are some key takeaways... The?International Medical Device Regulators Forum (IMDRF)?has finalized two key guidance documents that set expectations for?Software as a Medical Device (SaMD)?and?AI-enabled medical devices. --Takeaways-- 1. Enhanced Risk Characterization & Classification Clearer frameworks for?software risk assessment?and?device classification. Stronger?cybersecurity and interoperability?requirements. ??Tip:?Implement structured risk management early to ease regulatory approvals. 2. AI Transparency & Oversight Guidelines for?Generative AI and Software of Unknown Provenance (SOUP). Stricter rules for?dataset diversity and bias mitigation. ??Tip:?Ensure?AI traceability and explainability?to enhance clinician trust. 3. Continuous Monitoring & Post-Market Compliance New best practices for?real-world AI performance tracking. Clearer guidance on?AI model updates and cybersecurity protocols. ??Tip:?Establish?data monitoring frameworks?to improve patient safety. --What This Means for Developers-- Aligning with these guidelines ensures?faster regulatory approvals?and?stronger market positioning. Companies investing in?AI transparency, structured risk assessment, and post-market monitoring?will lead the industry. Read our blog post for more - https://lnkd.in/eHunzcni Reach out if you have questions or need help implementing these strategies. You can email us at [email protected] #medicaldevices #imdrf #regulation #regulatory #digitalhealth #ai #artificialintelligence #samd #radiology #remotemonitoring #digitaltherapeutics