St John Lynch, Niamh and Loughran, Roisin and Mc Hugh, Martin and McCaffery, Fergal (2026) Design and Expert Validation of an AI Change Management Process (AI-CMP) Assessment for AI-Enabled Medical Devices: Towards Agentic AI Change Management with Human Oversight. [Artefact]
![[thumbnail of AI-Agentic-Change Management Process v.4.0]](https://eprints.dkit.ie/style/images/fileicons/text_html.png) |
HTML (AI-Agentic-Change Management Process v.4.0)
- Submitted Version
Restricted to Repository staff only until 20 May 2028. Download (335kB) | Request a copy |
Abstract
Agentic Artificial Intelligence presents a significant opportunity to address one of the most resource-intensive obligations facing manufacturers of AI-enabled Medical Devices (AIeMDs): the assessment and documentation of iterative lifecycle changes. Recall evidence indicates that inadequately controlled modifications are a leading contributor to safety issues in AI/ML-enabled devices, yet change assessment remains largely manual. A prerequisite for safe agentic automation is a validated, structured instrument for the agent to execute. This paper reports the design and expert validation of the AI Change Management Process (AI-CMP) Assessment, a multi-jurisdictional instrument consolidating the change management expectations of EU MDR 2017/745, UK MDR 2002, FDA guidance, and the EU AI Act 2024/1689, together with standards including ISO 13485, ISO 14971, IEC 62304, IEC 81001-5-1 and IEC PAS 63621. In an exploratory, design-science study, the 35-section instrument was evaluated by a purposive panel of ten domain experts (eleven records, including one re-scored response) drawn from legal manufacturers and specialist consultancies across EU, UK and US regulatory contexts. Ratings indicated strong content validity for regulatory coverage and completeness, while lower ratings for length and resource burden strengthened the case for automation. Two implementations are reported: a proof-of-concept web application, and a deployed agentic prototype (AIa-CMP) operating under tiered, mandatory human oversight, applied in a real-world case study within a medical device organisation. These findings constitute purposive expert evaluation of content validity rather than population-level evidence of effectiveness; formal evaluation of the agentic implementation is identified as future work.
| Item Type: | Artefact |
|---|---|
| Additional Information: | Provided as a tool to MemoryTell Ltd on 20 May 2026 for use in change management. All rights reserved by N. St John Lynch and DkIT under Research Ireland funding as part of PhD. |
| Uncontrolled Keywords: | Change Management, AI Agentic, Co-Creation, Medical Device, SaMD, AIeMD, SiMD. |
| Subjects: | Business Computer Science Engineering Social Sciences > Law |
| Research Centres: | Regulated Software Research Centre |
| Depositing User: | Niamh StJohnLynch |
| Date Deposited: | 09 Sep 2026 12:28 |
| Last Modified: | 09 Sep 2026 12:28 |
| License: | Creative Commons: Attribution-Noncommercial-Share Alike 4.0 |
| URI: | https://eprints.dkit.ie/id/eprint/1104 |
Available Versions of this Item
-
AI-agentic Change Management Process (AIa-CMP) software tool built on Validated Change Management Process Framework v.4.0. (deposited 25 May 2026 10:22)
- Design and Expert Validation of an AI Change Management Process (AI-CMP) Assessment for AI-Enabled Medical Devices: Towards Agentic AI Change Management with Human Oversight. (deposited 09 Sep 2026 12:28) [Currently Displayed]
Actions (login required)
 |
View Item |
Downloads
Downloads per month over past year
