Validation of FDA-Regulated Medical Device and SaMD Products Using AI, ML, and LLMs, such as ChatGPT

The FDA is currently trying to modernize its systems and use of data. The FDA announced that all the agency’s centers must fully integrate generative AI into work by the end of June 2025. The intent is to reduce non-productive busy-work that has historically consumed the review process. Device submission review can often take many months, but there is an opportunity to reduce this timeframe and allow expert reviewers to focus on the more complex cases.

Medical devices using AI are designed to analyze vast amounts of data to generate clinical insights. This means that the company’s quality management system (QMS) must ensure consistent production and control of manufacturing and quality systems, and involves routine inspections and audits.
The Verifying Accurate Leading-Edge Development Act, or Valid Act, is pending and will codify the “firm-based” approach to regulation. The FDA will oversee methods used for technology development and validate reliability rather than decouple the AI product’s construction. By ensuring robust systems are in place, the FDA can enhance the overall safety and effectiveness of medical devices produced.

Rapid cycles of innovation inherent in products due to constant modification based on new information available pose challenges. ChatGPT from OpenAI has demonstrated substantial semantic medical knowledge and the ability to perform work that will accelerate the submission approval process.
Large Language Models (LLMs) trained on vast datasets embody the ultimate black box in the realm of FDA regulation. They are nonlinear and high-dimensional, making it difficult to trace specific inputs to outputs. A risk is that they may return wrong answers when trained on unreliable datasets. Under a firm-based regulation approach by the FDA, innovators can bring certain new products to market more efficiently.

LLMs will boost efficiency, but input data must be quality-checked. Industry must develop adequate standards and controls, evaluating AI algorithm models under the specific intended use of a device. Ultimately, the industry will be able to identify new product candidates and plan, execute, and analyze data from clinical trials.
In June 2025, the FDA announced plans to use AI to speed new medical device and SaMD approvals. Elsa is a tool that may enhance FDA review of safety data, summarize reports, and flag facilities needing inspection. Built within a high-security GovCloud environment, it offers a secure platform for FDA staff to access internal documents while ensuring information remains within the agency. Submission reviews could be considerably shortened.

Learn more about how FDA and life science companies are using AI and ChatGPT. This is expected to improve efficiency while enabling the FDA and companies to run more smoothly. This is critical at a time when we are faced with a rising demand for healthcare and physician shortages. Leveraging comprehensive data systems will lead to greater efficiency in diagnosis and treatment planning. Getting these products through the FDA regulatory submission process more quickly and efficiently is the goal to put them in the hands of patients.

For SaMD products, there are several guidance documents from FDA and the International Medical Device Regulators Forum that will be beneficial in understanding how to make changes to these products safely through a method using SaMD Pre-Specifications (SPS) and an Algorithm Change Protocol (ACP) to assess the potential risk and impact of a change.

Learn about the FDA’s AI/ML SaMD Action Plan, encouraging harmonization among technology developers on the development of GMLP. This is part of the FDA’s mission to define new policies and enable innovation while protecting public health. You will also learn about the Total Product Lifecycle (TPLC) Approach for SaMD regulation. This is a new paradigm focused on the assessment of an organization in terms of its software design, development, testing, and monitoring activities. This is part of the FDA’s Software Pre-Certification (Pre-Cert) Program. In some cases, a new 510(k) may not be needed, and documentation of change and analysis of risk management can take its place.

But this webinar doesn’t stop there!

We’ll provide an overview of computer system validation, including the draft guidance from the FDA on Computer Software Assurance (CSA), and the latest GAMP®5, 2nd Edition, that aligns with CSA. We’ll walk you through the Software Validation and Maintenance approach that will bring clarity to what the FDA is looking for, primarily allowing companies to manage risks from changes, while enabling improvement of performance and advancing patient care.

• Artificial Intelligence (AI) and Machine Learning (ML)
• Large Language Models (LLMs), such as ChatGPT
• GxP Systems
• Computer System Validation (CSV)
• Computer Software Assurance (CSA)
• Critical Thinking
• GAMP®5, 2nd Edition
• System Development Life Cycle (SDLC)
• GMP, GLP, GCP
• Validation Planning
• Requirements
• Risk Assessment and Mitigation
• Installation Qualification (IQ)
• Operational Qualification (OQ)
• Performance Qualification (PQ)
• Automated Testing
• Cloud Services
• Commercial-Off-the-Shelf (COTS) Software
• Software-as-a-Service (SaaS) Solutions
• Software-in-a-Medical Device (SiMD)
• Software-as-a-Medical Device (SaMD)
• FDA Compliance and Trends
• Industry Best Practices

• Information Technology (IT) Analysts
• IT Software Developers & Testers
• IT Support Staff
• IT Security Staff
• QC/QA Managers and Analysts
• Production Managers and Supervisors
• Supply Chain Managers and Supervisors
• Clinical Trial Data Managers and Scientists
• Compliance Managers and Auditors
• Lab Managers and Analysts
• Computer System Validation Specialists
• GMP, GLP, GCP Training Specialists
• Business Stakeholders using Computer Systems regulated by the FDA
• Regulatory Affairs and Submissions Personnel
• Consultants in the Life Sciences and Tobacco Industries
• Interns working at the companies listed above
• College students attending schools and studying computer system validation, regulatory affairs/matters (related to FDA), or any other discipline that involves adherence to FDA regulatory requirements.

Carolyn Troiano has more than 35 years of experience in computer system validation in the tobacco, pharmaceutical, medical device and other FDA-regulated industries. She has worked directly, or on a consulting basis, for many of the larger pharmaceutical and tobacco companies in the US and Europe. She is currently building an FDA computer system validation compliance strategy at a vapor company. Carolyn has participated in industry conferences, and is currently active in the Association of Information Technology Professionals (AITP), and Project Management Institute (PMI) chapters in the Richmond, VA area. Carolyn also volunteers for the PMI’s Educational Fund as a project management instructor for non-profit organizations.