Epic and AI in Swiss Hospitals: What's Already Here and What's Coming

Which Swiss Hospitals Are Already Using Epic?

Active Implementations

• LUKS Group – Live since 2019; Spital Nidwalden integrated early 2024
• Insel Gruppe / Inselspital Bern – Live since March 2024

Planned Deployments

• USZ Zurich – Contract awarded August 2025 (CHF 94.8M); earliest go-live 2027
• Kispi Zurich – Award November 2023 (up to CHF 51.2M); go-live by end of 2027
• CHUV + 11 Vaud hospitals – Contract awarded November 2025 following Federal Supreme Court dispute
• Canton of Bern – Legislative revision in public consultation

Key takeaway: Multiple Swiss university hospitals will operate Epic systems simultaneously by 2027.

What Epic's AI Can Already Do

Ambient Documentation

The system transcribes conversations during consultations and drafts clinical notes for physician review. Epic reports a meaningful reduction in documentation time, though independent verification remains limited.

Embedded Decision Support

The platform analyses labs, vitals, and medications in the background, surfacing relevant alerts: sepsis warnings, readmission risk, and treatment comparisons. This functionality draws from Cosmos, Epic's database containing roughly 300 million anonymised patient records and over 18 billion clinical encounters from over 300 health systems worldwide.

Critical caveat: The majority originates from US systems, raising questions about applicability to Swiss clinical guidelines and SwissDRG coding standards.

Administrative Tools

Penny, Epic's revenue cycle agent, suggests diagnostic codes, drafts insurance appeals, and automates claims follow-up — currently oriented toward US markets.

Three Emerging AI Assistants

Art (Clinician-Focused)

Prepares contextual information and writes clinical notes from conversations using AI Charting technology (developed with Microsoft). Available in the US from early 2026; Swiss rollout timing unannounced.

Emmie (Patient-Facing)

A MyChart-integrated chatbot explaining test results and suggesting appointments. While some Swiss institutions operate Epic-based patient portals, adoption here remains limited compared to US usage.

Curiosity (Prognostic Model)

Simulates disease trajectories to assess admission necessity, length of stay, and 30-day readmission risk. Trained on over 100 billion clinical events (developed with Yale and Microsoft). Pilot deployments expected 2026; Swiss clinical implementation likely years away.

Implications for Swiss Healthcare

LUKS and Inselspital automatically receive AI updates through Epic's biannual releases. By May 2024, Inselspital physicians accessed LUKS patient records directly within the system — marking the first live inter-institutional Epic data exchange in Switzerland.

Institutions still planning implementation (USZ, Kispi, CHUV) should clarify AI feature roadmaps before institutional decisions exclude clinical input.

Data Privacy Considerations

The US Cloud Act permits American authorities to request data from US companies regardless of server location. Zurich cantonal council members have filed multiple interpellations on this issue. Both USZ and LUKS emphasise Swiss server storage, though all affected hospitals procured Epic with awareness of this legal framework. The strategic question involves contractual safeguards rather than acceptance of inherent risk.

Conclusion

Effective AI implementation depends on models designed for local contexts. Healthcare professionals working in Epic institutions should verify which features are currently activated.

SwissMed AI welcomes field perspectives at hello@mdai.ch.