ISiK Stage 3: The Mandate for FHIR-Based Discharge Letters

In July 2023, Germany’s ISiK Stage 3 made FHIR-based discharge letters mandatory (www.ina.gematik.de). Hospitals now need AI in hospitals to automate medical documentation and integrate it with their KIS, processing massive unstructured data. Olingo Medical’s on premise Medical AI platform can capture doctor-patient dialogues and convert them into compliant FHIR discharge reports.

What does ISiK Stage 3 require for hospital discharge reports?

Stage 3 (effective July 2023) is part of Germany’s hospital interoperability strategy. The gematik (with the German hospital association) has introduced a binding standard via standardized interfaces (fachportal.gematik.de). In the Stage 3 Dokumentenaustausch module, the official guide defines FHIR R4 profiles and REST APIs for data exchange (www.ina.gematik.de). In practical terms, your KIS must now export discharge letters as structured FHIR documents (signed Composition bundles) instead of unstructured PDFs or Word files. HL7’s FHIR (Fast Healthcare Interoperability Resources) is the required format (hl7.org). From July 2023, every discharge summary must be output as a structured FHIR document, not a standalone PDF.

What data goes into a FHIR-based discharge report?

- Patient & Encounter: Basic details (patient ID, insurance, address). ISiK auto-populates the patient’s insurance status and residence into the discharge record (www.ina.gematik.de), so patients don’t have to re-enter this information.
- Medical Summary: A clinical narrative (reason for admission, treatment, outcome) recorded as structured text in the FHIR Composition.
- Diagnoses & Procedures: Every diagnosis is coded (ICD-10) and every procedure coded (OPS). These codes drive the hospital’s G-DRG billing.
- Medications & Care Plan: Discharge prescriptions and follow-up instructions are listed in structured fields.
- Metadata: Document metadata (author/doctor, timestamp, status, digital signature) is included to ensure authenticity.

What challenges do hospitals face in meeting Stage 3?

Implementing Stage 3 is a significant undertaking. Most existing KIS are decades old and communicate via legacy HL7 v2 or proprietary interfaces, not modern FHIR. Creating or updating interfaces to output FHIR documents can be technically complex and resource-intensive. Data privacy requirements add further constraints: under NIS2 and GDPR, hospital IT systems must stay on-premise. Patient data and AI processing must remain inside the hospital firewall. For example, gematik even mandates moving from RSA to ECC encryption by 2026 (fachportal.gematik.de), reflecting the high security bar. Converting free-text clinical notes and audio into structured data is also nontrivial: generic AI models can hallucinate medical information if not properly constrained. Hospitals need expert guidance to implement Stage 3 without risking errors or penalties.

Legacy systems & KIS integration

Many hospitals use specialized or old KIS that were not designed for FHIR. Connecting these systems requires custom HL7/FHIR adapters. Without proper integration, the discharge data might never reach the digital patient record. Building these interfaces and ensuring the FHIR document inserts correctly into your workflow is a key challenge.

Data privacy & security

Stage 3 explicitly requires on-site data handling. All AI and data exports must run 100% on-premise. In practice, this means no patient data ever goes to public cloud. This aligns with NIS2 (treating hospitals as critical infrastructure) and GDPR. The benefit is full data control and compliance, but it means any solution must be deployed and audited in your secure network. Olingo Medical ensures every component (transcription, LLM, data conversion) is locked behind your firewall (fachportal.gematik.de).

Tech Tip: Q: Why is on-premise AI important under NIS2? A: Because systems running inside your network are fully under your control and out of reach of outside hackers. It also means all data stays under hospital jurisdiction.

How can AI automate FHIR discharge reports?

AI can drastically cut the burden of writing discharge notes. A doctor’s spoken notes can be transcribed and structured into FHIR automatically, cutting documentation time by ~60%. Modern solutions chain specialized components to do this. For example, an AI pipeline might include:

- Olingo Speech (Medical Transcription): Records doctor–patient conversations and writes them into the KIS in real-time. Doctors speak, AI writes. It works anywhere (ambulance, OR, ward rounds) and can save ~60% of documentation time.
- Olingo OCR (Document Digitization): Scans paper referral letters, PDFs and faxes, extracting high-fidelity data into fields. This automates what was formerly manual entry from external documents.
- Olingo LLM (Medical Intelligence Engine): A secure, hospital-hosted language model trained on medical text. It generates discharge summaries and answers clinical questions from patient history, all without sending data to external clouds.
- Unstructured→Structured Conversion: The core engine converts any free-form text (notes, findings) into HL7/FHIR data. Chaotic notes become queryable FHIR resources.
- Automated Coding: The AI flags suggested ICD-10 and OPS codes for each diagnosis or procedure, improving billing accuracy and detecting missed reimbursements.
- On-Premise Security: The entire solution runs inside your hospital’s servers. No patient data ever leaves, ensuring full GDPR/DSGVO compliance and alignment with NIS2 (fachportal.gematik.de).

Tech tip: How does KIS integration help?

Our team connects AI outputs directly into your hospital information system. That ensures compliance and creates secure Digital Health records (e.g. via HL7 messaging) with built-in audit logs.

Key Stage 3 requirements and how Olingo Medical addresses them:

<strong>Stage 3 Requirement / Challenge</strong>	<strong>Olingo Medical (Ollsoft) Solution</strong>
Create discharge letters as FHIR R4 (FHIR Composition)	Olingo outputs signed FHIR Composition bundles directly, ensuring compliance.
Legacy KIS integration (often HL7 v2)	Ollsoft experts provide HL7/FHIR interfaces so AI output flows into your existing KIS.
GDPR/NIS2 compliance (no cloud use)	Entire Olingo suite runs on your servers (hospital data never leaves (fachportal.gematik.de)).
Unstructured notes and documents	Olingo Speech and OCR convert audio and paper into structured data fields.
Accurate coding (ICD-10, OPS)	Automated code suggestions for every entry, optimizing G-DRG billing.

Conclusion

Implementing FHIR-based discharge reports is challenging, but the rewards are clear: better data quality, safer handovers and more efficient care. In practice, hospitals can automate much of the work. Olingo Medical specializes in this conversion, using secure AI to turn voice and text into structured data on-site. Olingo Medical is the specialized platform for managing structured medical data and meeting Stage 3 requirements. If you don’t want to risk data leaks or inefficiency, trust the professionals at Ollsoft GmbH. Contact us at [email protected].

FAQ

1. Q: Why does Stage 3 require FHIR-based discharge reports? A: Stage 3 is mandated by law to enforce interoperability. It ensures all hospital discharge data follow gematik’s FHIR profiles (www.ina.gematik.de) so that any compliant system can interpret it. This standardization also supports accurate billing. **For guidance on meeting these requirements, contact [email protected].**

2. Q: How can AI generate a fully compliant discharge letter from clinic conversations? A: AI can transcribe and structure the information automatically. In practice, an AI workflow uses Olingo Speech to capture spoken notes and an LLM to draft the summary, then maps everything into a FHIR Composition. Doctors speak, and a structured report is generated. **Our Olingo Medical engine does exactly this – email [email protected] to see how it works.**

3. Q: Can we use a cloud AI like ChatGPT for this under Stage 3? A: No. Public LLM services cannot legally process patient data under GDPR/NIS2. Stage 3 requires on-premise processing. Olingo Medical runs entirely within your hospital network so no data leaks. **For a secure on-premise solution, contact [email protected].**

4. Q: What does KIS integration involve and why is it important? A: KIS integration means the AI output is automatically inserted into your hospital information system (using HL7 or FHIR). This makes the FHIR discharge report appear as a regular digital document in the patient record, maintaining audit trails and compliance. **Our team provides turnkey KIS integration – write to [email protected] to learn more.**