变革管理中的人工智能：医疗设备制造和注册中数据驱动技术的关键推动者

AI in Change Management: A KeyEnabler for Data-Driven Tech inMedical Device Manufacturing ANUSHA GANGADHARA, Associate Director, Product Owner, SmartSolve Quality and Regulatory Solution, IQVIADR. SANJA MATERN, Director, Regulatory Affairs, Regulatory Information Managemen t, Fresenius Medical Care Table of contents Introduction1Change requests: A recurring regulatory challenge2AI in change controls across QARA3A layered approach: Architecture and dynamic data handling4AI as a catalyst for data-driven technology4AI operating model: Process + people + governance5Process and reference data enhancements5People and skills5Governance and controls5Taking MedTech change control further with AI5Operational challenges with AI implementation5Conclusion6About the authors7 Introduction MedTech companies including medical devices,in vitrodiagnostics and softwareapplications that support and enhance patient care operate under intenseregulatory scrutiny while racing to innovate. Product and quality changes, Addressing such inquiries requires careful coordinationand adherence to regulatory processes, making quickturnarounds challenging despite the urgency expressedby stakeholders. The rigorous frameworks include majorregulations and standards formulated by FDA 21 CFR,EU MDR/IVDR, QMSR and ISO 13485, ISO 14971 and Change requests: A recurringregulatory challenge ‘A Saudi customer who attended the conference in the U.S.wants to know when the latest AI feature can be added totheir diagnostic equipment’…..’The pediatric extension ofthe adult screener has strong demand in India — what is Sounds familiar, right? Regulatory teams are frequentlyconfronted with targeted requests — almost on aquarterly or even monthly basis — for updates and Decision latency: Scattered data, lengthy access and interpretation times, and long approval cyclesrequiring manual intervention. Many organizations still rely on spreadsheets, emails, and static Data and functional silos: Engineering, quality, regulatory affairs, manufacturing, and clinical teams Documentation burden:High‑volume, complex changes, particularly for software and AI‑drivenconnected devices, span design files and records (DHF/DMR/DHR), Technical Documentation (TD), and Regulatory risk blind spots:With incomplete impact analyses done in fragmented sessions, regulatoryuncertainty to determine whether a change requires notification, submission, or approval by regulatory Cost of non‑compliance:Including observations, warning letters, rework, and recalls, driven by limitedpredictive capabilities in traditional change management, which is reactive and focuses on managing Artificial Intelligence (AI) could prove to be a practical accelerator for this discipline, addressing these pain points bymaking change management more data-driven, predictive and automated. Applied thoughtfully, AI augments change streamline and enhance change management withinregulatory and quality assurance (QARA) systems. Eachfeature targets a specific challenge in the traditionalprocess, illustrating how advanced technologies can AI in change controlsacross QARA Change management in MedTech refers to thesystematic process of proposing, evaluating, approving,implementing, and verifying changes that may affect Artificial intelligence can further transform manufacturing and quality processes by enabling real-time visibility,traceability, and knowledge reuse (herd wisdom). In manufacturing and quality, AI can be used to optimize processes,triage deviations, and evaluate supplier changes to improve efficiency and compliance. For software-related products, AI as a catalyst fordata-driven technology A layered approach:Architecture and dynamic AI-driven architecture and data handling enhancecompliance and efficiency in manufacturing andquality processes by automating tasks, improving dataintegration, and supporting regulatory requirements. Key components include a data ingestion layer thatconnects to and refreshes live regulatory releasesworldwide to ensure the most current compliancerequirements are available for processing. LayeredAI services drawing from this data layer support Real-time visibility Scalable Artificial intelligence improvestraceability by automaticallyconnecting related records andproviding clear explanations for AI models leverage historicalchange data to identify effectivestrategies and standardizepractices, creating an AI-powered dashboards unifyquality, manufacturing, andregulatory metrics, enablingfaster and more informed 2. Generative regulatory intelligence is anotherimportant concept to enable dynamic gap analysis AI operating model: Process +people + governance Process and reference data enhancements Logical standardization and the development of a strongregulatory backbone across markets are essential. Thisincludes tiered classification frameworks for changes 3. Continuous post-market feedback loops integratereal-world data and service logs to keep riskassessment