A proposed reference model for the deployment of an integrated ai system in a large oncology center under the EU AI Act and MDR. Part I: Strategic & Operational Framework
Andrzej Jankowski
UWMDominik Wawrzuta
Mateusz Dąbkowski
Ewelina Żarłok
Lech Polkowski
Andrzej Skowron
Piotr Artiemjew
Abstrakt
Large-scale deployment of AI in oncology is constrained less by standalone algorithmic performance than by system-level safety, accountability, interoperability, and regulation-aware governance. Grounded in approximately one year of practical pre-deployment work within the OnkoBot project, this paper specifies a deployment- and governance-first reference model for integrated oncology AI platforms under the EU AI Act and the Medical Device Regulation (MDR).
The paper introduces Architecture for Medical AI Collaboration (AMAC), an implementation-neutral, system-level envelope that enforces strict online/offline separation between clinical operation and model/knowledge learning and evolution, gate-controlled releases via a Clinical Governance Gateway (CGG) with explicit human-in-the-loop (HITL) escalation, and tamper-evident auditability across clinical, technical, and interoperability boundaries. AMAC is anchored by the Community of Collaborative Evolving Medical Assistants (CEMA), a supervised multi-agent computational core that performs coordinated clinical reasoning under bounded autonomy.
Concrete deliverables include: (i) a reference architecture outline with explicit responsibilities and auditable control points; (ii) a phase-gated deployment pathway (Preparation → Prototype → Pilot → Integration → AMAC operation) with required evidence packs, decision gates, and rollback/suspension mechanisms; and (iii) enforceable socio-technical gate criteria, including Socio-Technical Readiness Levels (STRL), readiness metrics, and accountability mapping (RACI). The model is intentionally non-normative and does not encode clinical guidelines; it provides a minimal, auditable governance architecture designed to make large-scale clinical AI integration feasible, controllable, and regulation-compatible in complex oncology environments.
Słowa kluczowe:
oncology, integrated AI platform, reference architecture, reference deployment pathway, AI governance, auditability, EU AI Act, MDR, SaMD, human-in-the-loop (HITL), retrieval-augmented generation (RAG), GraphRAGBibliografia
Barredo Arrieta A., Díaz-Rodríguez N., Del Ser J., Bennetot A., Tabik S., Bar-Bado A., García S., Gil-López S., Molina D., Benjamins R., Chatila R., Her-Rera F. 2020. Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Information Fusion, 58: 82-115. Retrieved from https://www.sciencedirect.com/science/article/pii/S1566253519308103 (21.12.2025). https://doi.org/10.1016/j.inffus.2019.12.012 Google Scholar
Bender D., Sartipi K. 2013. HL7 FHIR: An Agile and RESTful approach to healthcare information exchange. Proceedings of the 26th IEEE International Symposium on Computer-Based Medical Systems (CBMS): 326-331. https://doi.org/10.1109/CBMS.2013.6627810 Google Scholar
Chang E.Y. 2025. Multi-LLM Agent Collaborative Intelligence: The Path to Artificial General Intelligence. ACM Books. https://doi.org/10.1145/3749421 Google Scholar
Dąbkowski M., Wawrzuta D., Żarłok E., Jankowski A., Polkowski L., Skowron A., Artiemjew P. 2025. OnkoBot: Propozycja Karty Projektu. Projekt Zintegrowanego Systemu AI dla Narodowego Instytutu Onkologii PIB. Internal project document (NIO-PIB and UWM), Warsaw/Olsztyn, version dated 5 October 2025. Available upon request from the authors (internal circulation). Google Scholar
European Parliament and Council. 2016. Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data (General Data Protection Regulation – GDPR). Official Journal of the European Union, L 119. Google Scholar
European Parliament and Council. 2017. Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices (MDR). Official Journal of the European Union. Retrieved from https://eur-lex.europa.eu/eli/reg/2017/ 745/oj/eng (21.12.2025). Google Scholar
European Parliament and Council. 2024. Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal of the European Union, L series, 2024/1689, 12 July 2024. Retrieved from http://data.europa.eu/eli/reg/ 2024/1689/oj (21.12.2025). Google Scholar
Filipovic K., Said R., Saban L., Uzdilli M., Krey M. 2026. Toward an AI Maturity Model in Healthcare: Identifying Core Dimensions and Critical Success Factors. Proceedings of the 59th Hawaii International Conference on System Sciences (HICSS-59): 6619-6633. Google Scholar
Gerke S., Minssen T., Cohen G. 2020. Ethical and legal challenges of artificial intelligence-driven healthcare. Artificial Intelligence in Healthcare: 295-336. https://doi.org/10.1016/B978-0-12-818438-7.00012-5 Google Scholar
Holzinger A., Langs G., Denk H., Zatloukal K., Müller H. 2019. Causability and explainability of artificial intelligence in medicine. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 9(4): e1312. https://doi.org/10.1002/widm.1312 Google Scholar
Institute for AI Industry Research (AIR), Tsinghua University. AIR. 2024. creates a virtual hospital, enabling AI doctors to self-evolve. Web publication, 24 May 2024. Retrieved from https://air.tsinghua.edu.cn/en/info/1007/1872.htm (21.12.2025). Google Scholar
International Electrotechnical Commission. 2015. *IEC 62304:2006+AMD1:2015 CSV – Medical device software – Software life cycle processes. Google Scholar
International Organization for Standardization. 2019. ISO 14971:2019 – Medical devices – Application of risk management to medical devices. Google Scholar
International Organization for Standardization. 2016. ISO 13485:2016 – Medical devices – Quality management systems – Requirements for regulatory purposes. Google Scholar
International Organization for Standardization. 2013. *ISO/IEC 27001:2013 – Information security management systems – Requirements. Google Scholar
International Organization for Standardization. 2016. *ISO 27799:2016 – Health informatics – Information security management in health using ISO/IEC 27002. Google Scholar
Jankowski A. 2017. Interactive Granular Computations in Networks and Systems Engineering: A Practical Perspective. Springer. Google Scholar
Jiang L., Wu Z., Xu X., Zhan Y., Jin X., Wang L., Qiu M. 2021. Opportunities and challenges of artificial intelligence in the medical field: Current application, emerging problems, and problem-solving strategies. Journal of International Medical Research, 49(3). https://doi.org/10.1177/03000605211000157 Google Scholar
Li J., Lai Y., Li W., Ren J., Zhang M., Kang X., Wang S., Li P., Zhang Y.-Q., Ma W., Liu Y. 2024. Agent Hospital: A Simulacrum of Hospital with Evolvable Medical Agents. arXiv:2405.02957. Retrieved from https://arxiv.org/abs/2405.02957 (21.12.2025). Google Scholar
Mandel J.C., Kreda D.A., Mandl K.D., Kohane I.S., Ramoni R.B. 2016. SMART on FHIR: A standards-based, interoperable apps platform for electronic health records. Journal of the American Medical Informatics Association, 23(5): 899-908. https://doi.org/10.1093/jamia/ocv189 Google Scholar
Pedrycz W., Skowron A., Kreinovich V. 2008. Handbook of Granular Computing. Wiley, Hoboken. Retrieved from https://onlinelibrary. wiley.com/doi/book/10.1002/9780470724163 (21.12.2025). https://doi.org/10.1002/9780470724163 Google Scholar
Polkowski L. 2009. Granulation of knowledge: similarity based approach in information and decision systems. In: Encyclopedia of Complexity and Systems Science. Springer. Google Scholar
Price W.N., Gerke S., Cohen I.G. 2019. Potential liability for physicians using artificial intelligence. JAMA, 322(18): 1765-1766. https://doi.org/10.1001/jama.2019.15064 Google Scholar
Skowron A., Jankowski A., Dutta S. 2025. Interactive Granular Computing: Toward Computing Model for Complex Intelligent Systems. Proceedings of the 20th Conference on Computer Science and Intelligence Systems (FedCSIS): 59-72. Retrieved from https://annals-csis.org/Volume_43/drp/pdf/6355.pdf (21.12.2025). Google Scholar
Yang J., Soltan A.A.S., Eyre D.W., Yang Y., Clifton D.A. 2022. Human-in-the-loop machine learning in clinical oncology: challenges and opportunities. The Lancet Digital Health, 4(10): e706-e708. https://doi.org/10.1016/S2589-7500(22)00145-3 Google Scholar
UWM
