Reaching the last mile? How Design Decisions Shape Medical Device Qualification and Classification of Patient Support Chatbots In The EU
A design-oriented regulatory analysis of a chatbot supporting stroke survivors, in which HCI researchers and legal scholars use Design Space Analysis (the Questions, Options, Criteria framework) to trace how variations in three conversational functions, health question answering, recovery monitoring and motivating medical adherence, change qualification and classification under the EU Medical Device Regulation, the GDPR and the AI Act. Seemingly minor design choices such as role, actionability and proactivity move regulatory obligations; the paper sets out ten conversational design considerations and compliance-aware pathways to market.
Publisher
ACM (Proceedings of the 8th ACM Conference on Conversational User Interfaces, CUI 2026); Aalborg University; Masaryk University; Health Management Institute Brno
Published
21 Jul 2026
Added
today
Key Findings
- Three conversational functions of a stroke-survivor support chatbot (health Q&A, recovery monitoring, adherence motivation) were varied through the QOC framework and each variation assessed against the MDR, the GDPR and the AI Act.
- Design decisions about the chatbot's role, the actionability of its output and its proactivity change whether and how it qualifies as a medical device and what obligations follow.
- The paper contributes ten conversational design considerations that influence regulatory interpretation and shows how QOC can carry legal considerations into human-centred design workflows.
- It proposes compliance-aware pathways to market and asks whether every chatbot should attempt the 'last mile' into clinical practice.
Methodology Notes
Interdisciplinary case analysis (HCI and law) on a single design case, no user study; Horizon Europe funded. Published 2026-07-21 in the CUI 2026 proceedings (pp. 1 to 17, CC BY 4.0). Verification via Crossref metadata (title, authors, affiliations, date, license) and the OpenAlex deposited abstract; dl.acm.org is unreachable from this workspace and the Aalborg repository copy blocks file downloads, so the ten considerations are not listed here.
Topics
Authors
Milo Marsfeldt Skovfoged, Frantisek Kasl, Michal Koščík, Pavel Loutocky, Vaclav Stupka, Robert Mikulik, Hendrik Knoche
Tags
Cite This
APA
Milo Marsfeldt Skovfoged et al. (2026). Reaching the last mile? How Design Decisions Shape Medical Device Qualification and Classification of Patient Support Chatbots In The EU. ACM (Proceedings of the 8th ACM Conference on Conversational User Interfaces, CUI 2026); Aalborg University; Masaryk University; Health Management Institute Brno. https://doi.org/10.1145/3816046.3816232