zurück zum Artikel

US providers of monolithic software like Epic Systems in hospitals endanger digital sovereignty and threaten to become a cost trap.

Hospital information systems like Epic from Epic Systems offer fast integration, standardized processes, and a solid foundation for AI. At the same time, they create dependencies and limit flexibility. Hospital software like Epic was recently discussed at the Digital Health Innovation Forum of the Hasso Plattner Institute.

Jorge Juan Fernández, Chief Innovation Officer at Hospital Clínic de Barcelona, placed Epic in the context of different global models in his presentation “The European Third Way: Examples and lessons in re-engineering our healthcare systems through European innovation to drive shared prosperity”, describing the platform as a “monolithic stack” that enables much but does not allow everything. At the same time, he pointed out that Epic has long been in use in many large healthcare systems, especially in the US, and is increasingly being introduced or planned in Europe.

Epic becomes a billion-dollar question

Germany is also considered an example of this trend. As the trade publication kma reports [1], following a legislative amendment, the Berlin Charité will be allowed to take out loans of up to 400 million euros by 2035, among other things, to finance the use of Epic, [2] the introduction of which alone is estimated at around 200 million euros. Until now, Charité relied on SAP, which had announced its withdrawal from the HIS market.

Fernández contrasted this with an alternative model: open platforms based on standards such as openEHR, an open international standard for electronic patient records (Electronic Health Record, EHR), which enables standardized storage, management, and exchange of health data. These approaches offer more flexibility and interoperability but also require significantly more coordination effort. Fernández assumes that these two models will diverge in the coming years.

As part of his “European Third Way,” he advocated for Europe to develop its own approach between the US model – characterized by finance-driven, technology-centric hospital care – and the Chinese model, which focuses heavily on scaling and industrial efficiency. Europe should focus more on prevention, primary care, and the integration of health and social data.

Epic and AI

Chenzhe Cao, Principal Technology Strategist at Mass General Brigham, described Epic in his presentation “Operationalizing Intelligence: A Systems Approach to AI Enablement for an Integrated Academic Health System” as the central infrastructure of his clinic network. „Epic is very much a transactional system“. No structured documentation, no billing; no data, no processes. It is precisely this integration that makes it possible to bring AI applications into clinical practice on a large scale.

In the context of AI, this is a decisive advantage because data is available in large quantities and in a consistent structure. At the same time, Cao pointed to the economic pressure in the US system: many clinics are operating close to or below the profitability limit; the situation is similar in Germany [3]. According to Cao, AI must now be integrated into care to overcome the economic and personnel problems of healthcare systems at all. The key factor here is not only the technology but its embedding in clinical processes, clear governance structures, and the involvement of staff. AI should support the work of doctors and nurses, not replace it.

The example of German hospital operators clearly showed how different the starting positions are in Europe. Representatives of large hospital groups described historically grown IT landscapes with multiple parallel core systems and hundreds to thousands of subsystems, which are often rarely interoperable. Instead of a radical system change, many facilities therefore rely on an additional data layer – a kind of “clinical data layer” that aggregates information from various sources. Ähnliche Ansätze verfolgt beispielsweise auch die Software von Palantir [4].

The example of Denmark provided further insight. Malene Fischer, Deputy Chief Executive for Research, Innovation & Education at Rigshospitalet in Copenhagen, described the introduction of Epic as a conflict-ridden process. Looking back, she described the initial phase as “approaching civil war”. The background was massive changes in everyday hospital life and considerable resistance from staff.

Today, Rigshospitalet follows a more pragmatic approach. Epic serves as the core system, while innovation often takes place in parallel systems. New applications, especially in the AI field, are developed in separate environments, access the data, and feed results back. This creates a hybrid architecture. Particularly in the field of AI, the discussion showed that the greatest effects are not necessarily in classical medicine but in organization. Data-based forecasts allow resources to be deployed more effectively and processes to be improved.

Value of health data

Significant differences were also apparent in the area of data usage. Michael Hübner, responsible for innovation and digitalization at Sana Kliniken, emphasized that while health data has considerable economic value. “The world needs to pay for data”, he said. its direct sale is not legally permissible in Germany and is culturally difficult. Instead, models are emerging that are legally permissible and “ethically in a very good way” and help “promote better healthcare”, in which data is used in cooperation with research and industry to both promote innovation and finance one's own data infrastructure, a “win-win situation.”

Discussion on risks and benefits of innovation called for

Ian Abbs, Strategic Advisor at Guy’s and St. Thomas’ NHS Foundation Trust, criticized a misalignment in dealing with risks. The risk of doing nothing is completely ignored or not taken seriously. Therefore, “a much more open and transparent conversation” about the risks and benefits of innovation is needed.

Abbs also referred to political initiatives such as the “10-year plan for the UK health system,” in which technological innovation and data infrastructures are to play a central role. The prerequisite for this is large, secure data spaces that are used jointly with regulatory authorities to introduce innovation in a controlled manner. The use of Palantir in the English healthcare system [5] or cyber incidents were not discussed [6].

Timo Minssen, Professor at the University of Copenhagen, emphasized that innovation in healthcare has historically never arisen from complete security but from controlled experimentation: “innovation in medicine has never been achieved by freezing practice,” he said. The goal is to test new technologies under supervision and gradually implement them.

He referred to international examples such as in Singapore and an AI real-world laboratory in Utah. There, the chatbot from the start-up Doctronic is currently being tested [7] to automatically extend follow-up prescriptions according to defined guidelines; however, security experts have already pointed out weaknesses [8]. In Germany too, more and more AI real-world laboratories are emerging, for example in Baden-Württemberg [9].

“Europe must hurry”

Minssen stressed that there is “clearly need for speed” and that Europe risks falling behind in the international competition for digital technologies. At the same time, the real-world laboratories must be designed robustly, being flexible in processes but clear regarding security, values, and liability.

Organizations are currently working on guidelines and standards for the use of AI in healthcare. This also involves coordinating such approaches not only nationally but across Europe.

Bettina Goerner, Chief Data Officer of the cloud-based practice software Eterno, also pointed to the growing regulatory complexity and spoke of a “perfect storm” of overlapping regulations – from the GDPR to the MDR and the AI Act, whose requirements can be in conflict, according to legal experts [10]. Companies therefore need not only rules, but above all clear interpretative aids.

Thomas Renner from the Federal Ministry of Health referred to new funding programs, including around 650 million euros for the next ten years for data infrastructures and regulatory “sandboxes” in Germany. The goal is to improve data access, structure regulatory processes, and bring innovation into application faster.

[11]

(mack [12])

URL dieses Artikels:
https://www.heise.de/-11229145

Links in diesem Artikel:
[1] https://www.kma-online.de/aktuelles/politik/detail/novelle-universitaetsmedizingesetz-charite-soll-selbst-kredite-aufnehmen-fuer-epic-kis-55236
[2] https://www.heise.de/news/Epic-Systems-US-Anbieter-gewinnt-Ausschreibung-der-Charite-11114030.html?from-en=1
[3] https://www.heise.de/hintergrund/Interview-zur-Krankenhausreform-Komplexitaet-der-Datenmeldungen-nimmt-stark-zu-10324441.html?from-en=1
[4] https://www.heise.de/news/Palantir-Chef-Karp-Wir-unterstuetzen-Kriegsfuehrung-und-wir-sind-stolz-darauf-11211101.html?from-en=1
[5] https://www.heise.de/news/Palantir-im-NHS-England-Sorge-vor-Datenmacht-und-politischem-Missbrauch-11218642.html?from-en=1
[6] https://www.heise.de/news/Cyberangriff-auf-Londoner-Labordienstleister-Kliniken-benoetigen-Blutspenden-9759298.html?from-en=1
[7] https://commerce.utah.gov/ai/agreements/doctronic/
[8] https://mindgard.ai/blog/doctronic-is-now-accepting-new-patients-and-unsafe-instructions
[9] https://www.heise.de/hintergrund/Interview-Datenschutz-schuetzt-die-Schwaecheren-gegenueber-maechtigen-Akteuren-11141922.html?from-en=1
[10] https://www.heise.de/hintergrund/eHealth-Interview-KI-Verordnung-und-DSGVO-Risiken-und-Co-9327885.html?from-en=1
[11] https://www.heise.de/newsletter/anmeldung.html?id=ki-update&wt_mc=intern.red.ho.ho_nl_ki.ho.markenbanner.markenbanner
[12] mailto:mack@heise.de
[13] https://www.facebook.com/heiseonlineEnglish
[14] https://www.linkedin.com/company/104691972
[15] https://social.heise.de/@heiseonlineenglish
[16] https://www.heise.de/news/Digital-Health-Diskussion-um-Epic-Systems-offene-Plattformen-und-weitere-Wege-11228964.html

Copyright © 2026 Heise Medien