Listen to the podcast below:
FDA Sets Rules Of The Road For AI Device PCCPs
The US Food and Drug Administration has released a final guidance on pre-determined change control plans (PCCPs) for devices using AI, highlighting rapid changes in health care AI since the draft was published 18 months ago.
PCCPs allow manufacturers to get prior approval from the agency for anticipated future changes at the time of the initial regulatory clearance. The option is especially valuable for manufacturers of software-driven devices that must be regularly updated for security and performance.
While the final version is similar to the draft, it includes a new section that discusses version control and maintenance for device PCCPs. Some language has also been edited for clarity: for example, while the draft spoke of Artificial Intelligence/Machine Learning, the final version uses a more general term, AI-Enabled Device Software Functions (AI-DSFs).
The guidance document explains that PCCPs for AI-related changes should include three components: a detailed description of the proposed modifications, a modification protocol that explains how changes will be validated, and an impact assessment that looks at the benefits and risks of each modification. An appendix to the document includes sample PCCPs for a range of devices.
Penetration Testing Helps Ensure Device Security
While cybersecurity has been a hot topic for years, one specific component – penetration testing – has been coming to the forefront recently. In penetration testing, a “white hat” hacker tries to find any vulnerabilities in a device’s cybersecurity so issues can be fixed before they can be exploited by bad actors.
The process is especially significant for the European market, where notified bodies will often turn away device submissions that do not reflect penetration testing. In the US, it’s not strictly required but is strongly recommended for any device with a software or hardware component, even if the component is not expected to connect to the internet.
While manufacturers can do their own penetration testing, it’s typically not recommended due to the potential conflicts of interest. Instead, cybersecurity experts recommend looking for a testing company with a good reputation and experience in areas such as the Internet of Things and artificial intelligence, if the device incorporates it.
The importance of a strong cybersecurity program is expected to grow in the coming years due to new patient privacy and data security laws in the EU, as well as in US states like California and Texas.
Digital Health Advisory Committee Focuses On Generative AI In First Meeting
In October 2023, the FDA formed a new advisory committee tasked with advising the agency on digital health issues, reflecting the growing importance of digital technologies on the health care sector.
In November, that committee — the Digital Health Advisory Committee — held its first meeting in which it focused on the promise of generative AI-enabled medical devices as well as challenges the agency faces in reviewing them.
FDA Commissioner Robert Califf began the inaugural meeting emphasizing the enormous potential digital health technologies have in meeting critical health care issues, especially in addressing disparities in traditionally underserved groups. However, he noted the FDA must evolve to continue to oversee the safe and effective use of AI ensuring compliance while fostering innovation as the technology progresses.
Some of the specific topics the committee discussed included creating a regulatory framework for governing generative artificial intelligence, or GenAI-medical devices, and ensuring their safety and efficacy throughout their total product life cycle (TPLC).
Committee members also addressed three discussion questions concerning premarket performance evaluation, risk management, and postmarket performance monitoring.
The committee also touched on issues unique to AI, such as “hallucinations” — when AI generates fabricated information; and “drift” rates, which refers to changes in data that can degrade an AI system over time.
The committee also stressed the importance of keeping “humans in the loop” and highlighted the need to train clinicians and patients on using GenAI devices.
Deloitte 2025 Life Sciences Outlook Report Finds Medtech Leaders Focus on AI/GenAI
The Deloitte 2025 Life Sciences Outlook report, which surveyed 100 executives from the life sciences and 50 executives from medical device manufacturing companies across the US, Europe and Asia to learn about their concerns and priorities ahead, found that 75% are optimistic about the year ahead; about 60% of respondents see generative AI and digital transformation as “key emerging trends” and plan to increase investments in these areas across the value chain.
Digital transformation is expected to have a major impact on life sciences’ organizational strategies. Deloitte expects that medtech companies that leverage AI, particularly generative AI capabilities, could achieve 6% to 12% in cost-efficiencies in the next two to three years.
Medtech leaders reported that using AI and GenAI has generated the greatest value in product development, 42%, and cybersecurity, 35%. For product development, organizations leverage AI for concept design and prototyping and GenAI for summarizing scientific literature and drafting trial documents, such as vendor contracts and site checklists.
For IT, respondents said, they found GenAI useful to generate high-quality code and to improve data management processes, such as creating meta labels, cleaning data and anonymizing data to enhance security and useability, the report said.
Five AI Technologies Selected For UK MHRA’s AI Airlock Pilot
The UK’s Medicines and Healthcare products Regulatory Agency (MHRA) has selected five AI technologies for its AI Airlock pilot scheme, a regulatory sandbox scheme to help stakeholders understand the challenges of regulating AI in medical devices.
The MHRA launched the scheme in July and closed it for applications on 7 October. Lenus Health, Philips, Newton’s Tree, OncoFlow, and SmartGuideline are the successful companies
The Airlock is not a route to market and is not directly linked to product approval. The findings will be announced in 2025 and are expected to influence future UK AI medical device guidance.
The scheme supports the in-progress National Health Service’s (NHS) 10-year plan to revolutionize the delivery of care with digitally enabled tools. It also aims to address recommendations from Lord Ara Darzi’s NHS review that suggested a regulatory framework to foster AI innovation and safety.
Selected technologies include Lenus Health’s AI for predicting chronic obstructive pulmonary disease outcomes, Philips' AI for summarizing radiology reports, Newton’s Tree’s real-time AI performance monitoring, OncoFlow’s personalized breast cancer management plans, and SmartGuideline’s AI for simplifying access of information in NICE guidelines.
IMU Biosciences Joins £20M UK Government-Backed Cancer Immunoprofiling Project
The UK has launched a nationwide multi-million cancer immunotherapy project called Multiomic Analysis of Immunotherapy Features Evidencing Success and Toxicity (MANIFEST). The project is backed by over £20m (~$25m) in funding from government and industry and involves 16 UK academic institutions and 11 industry partners.
It aims to profile the immune systems of 6,000 patients with breast, bladder, kidney, and skin cancer. Half of the patients in the cohort have completed therapy, while the other half are about to start treatment.
IMU Biosciences, a London-based startup, is part of the consortium. The IMU platform profiles patients' immune systems to predict how variations in immune cells affect cancer treatment response. It uses full spectrum flow cytometry to classify immune cell subsets, providing a detailed view of the immune system.
Immunotherapy can cause side effects by taking the “brakes off the immune system,” Adam Laing, IMU co-founder and chief scientific officer, told Medtech Insight. In other words, the immune system can react to healthy tissue. A better understanding of how and why adverse reactions happen could transform how patients are cared for, said Laing.
