Architecture, Governance and Migration
of Clinical Data.
Over 30 years of experience processing and analyzing structured and unstructured clinical data. Specialized in interoperability, semantic integrity and sustainable data architectures — grounded in our own formal research (AION · SILD).
Inspection at
Data-boundaries
Clinical data can lose meaning in transfer.
Even when the transfer succeeds technically.
When a clinical finding is transferred from a hospital information system to a FHIR server, information can be lost — even when the transfer succeeded technically and the target validation passes. SILD compares source and target along four loss patterns from the FM-4 model. For today’s technical classification: statements about coverage presuppose a declared model and fixed decomposition and variation assumptions; one finding can involve several patterns at once. The result is traceable findings against agreed review rules — not a general proof of lossless transfer.
Inputs of every review: matched source and target data plus review rules agreed in writing — not the target data alone.
Four loss patterns from FM-4 — one example each:
ORC → basedOn)08:12–11:47 → 00:00–23:59 (all-day)V/A/Z/G)The cards show possible anomalies, not findings: a missing reference resolution, a missing side marker or an empty single field does not by itself prove information loss — the information may be preserved elsewhere, in another coding or with a converted unit. Whether agreed information is actually missing depends on the complete target context and the review rules. No clinical priorities follow from the pattern labels.
Three disciplines, one methodology.
Wherever data moves between systems, losses can occur — semantic, structural, operational. Our three service areas address exactly these transitions: from FHIR profile to platform migration, always with the claim of traceable review.
FHIR Governance & Semantic Integrity
Profiles, terminologies and mappings reliably maintained after go-live. This is not an implementation task — it is an operational discipline. "Go-live is a project. Operations is a discipline."
Clinical Data Architecture & Migration
Consolidate heterogeneous subsystems — based on our own CDV methodology (Clinical Data View) and three decades of database experience across eight systems. Hands-on, testable, documented.
Platform & Infrastructure
Scalable storage, compute and DevOps architectures for clinical and industrial applications — from ZFS NAS to object cloud, from PostgreSQL tuning to end-to-end DevOps pipelines.
Traceable review of a data transfer.
The SILD Pilot reviews one defined transfer source → target: up to 1,000 unambiguously matched source/target pairs from released test data or effectively anonymised data, against up to ten review rules agreed in writing — for example value and unit, code and code system, or time period. Paid pilot. Binding fixed price once the scope of review has been jointly defined. The 30-minute initial call is free of charge and without obligation.
Review agreement
Together we fix in writing: source data and target data, mapping, review rules, versions, responsible persons and a secure transfer channel. Feasibility is clarified before you commission the pilot.
Review run and report
One status per case and review rule — met, deviation evidenced, unresolved, not reviewable, not applicable — with evidence, denominator and review coverage, plus a machine-readable findings list. No global percentage.
Next steps
Prioritised technical next steps, assigned to the responsible teams for review, and an explanatory call. One limited re-test after an agreed correction is included.
AION & CAIRN — the formal grounding.
SILD is based on a formal model. The foundations — Allen relations, DAG-based type hierarchies, causal inference, ε-differential privacy — are published as their own scientific ecosystem: AION as the formal model, CAIRN as the Python reference implementation. Open Source under AGPL-3.0, as self-archived preprints on Zenodo (not yet peer-reviewed), available on PyPI.
AION
Formal model for clinical data: 13 Allen relations, DAG-based formal type system with model-related SMT checks (Z3), causal inference, explainable AI, ε-differential privacy.
aion-clinical.eu →CAIRN
Python reference implementation with bidirectional FHIR R4 mapping, HL7 v2 adapters, MLLP listener — and the SILD detector as integrated component. Current version aion-clinical 2.0.7.
ISCaD GmbH — Independent Software, Consulting and Development.
Clinical data architecture is a discipline — not an improvisation
Founded in 2000, based in Wedemark near Hannover, Germany. Specialized in architecture, governance and migration of clinical data — especially where the transition between systems becomes critical: FHIR profiling, semantic integrity, data migration.
Our own formal research in the AION / CAIRN / SILD ecosystem with eight self-archived preprints on Zenodo — not peer-reviewed. AGPL-3.0 open source. Commercial licenses on request.
- Founded
- 2000 · Wedemark, Lower Saxony, Germany
- Legal form
- GmbH · HRB 220407 · Local Court Hannover
- Management
- Friedhelm Matten
CV on LinkedIn → - Focus areas
- Clinical data · FHIR / HL7 / DICOM · Interoperability · Migration · Platform architecture
- Research
- AION · CAIRN · SILD · 8 preprints on Zenodo · FM-1 to FM-4.2 · determination of limits · scaling layer
- License
- AGPL-3.0 · Commercial on request
What we have built.
A selection of mandates that show methodology and depth of work. The order follows substantive focus, not chronology.
Clinical Information & Analytics System
Web-based analytics system for clinical data at Hannover Medical School. Integration of patient, surgical, heart-lung machine, DICOM and laboratory data via HL7 interfaces. Real-time data as ADT/ORU messages, continuous integration of historical legacy data.
HL7 Integration & Quality Assurance
Clinical data integration from HIS, HL7, iMedOne, heart-lung machine and legacy systems. Patient-level data streams supporting national and international quality assurance procedures in pediatric cardiology and pediatric cardiac surgery. Web application for automated QA data entry control and manual correction after release.
Clinical Data View (CDV)
Development of a data structure for managing and analyzing complex structured and unstructured clinical data. Migration tool for consolidating data stocks from subsystems with heterogeneous database structures and types into CDV — the methodological foundation of many of our mandates.
ZFS Server System / SOPHOS Firewall
Construction of a ZFS-based server system with distributed replication. Samba server with ACLs for mixed client environments (MacOS / Windows / Linux). SOPHOS firewall as network perimeter with segmented zones for operations, administration and backup. Subsequent migration of data to object storage (MinIO with S3 interface) and development of analytics and administration tools for ongoing operations.
Backend Platform & DB Optimization
Integration of kitchens and online order / delivery processes (Uber, Wolt, Lieferando). PostgreSQL performance tuning, DevOps setup (GitLab, WireGuard), stabilization of production-grade operational structures. Extensive Python development on the backend.
HealthSafe24 — CIO & Shareholder
Responsibility for technology and architecture at HealthSafe24 GmbH as Chief Information Officer and co-shareholder. Subsequently external advisory role for strategic and technical development.
A selection of past mandates.
Grouped by domain. A deliberately restrained selection from three decades — not all projects are listed here, some are under confidentiality agreements.
Healthcare · Science
- Medizinische Hochschule Hannover
- Herzzentrum Leipzig
- Herz- und Diabeteszentrum NRW
- HealthSafe24 GmbH
Public sector · Education
- Region Hannover
- Deutscher Industrie- und Handelskammertag
- Niedersächsisches Studieninstitut
- Technologie-Zentrum Hannover GmbH
Business · Industry
- Informix Deutschland GmbH
- Deutsche Telekom AG
- T Systems
- EXPO 2000 GmbH
- Media company (undisclosed)
- Cloudeatery GmbH
- Yobst GmbH
- Nicolai Sanitätshäuser
What we think and write.
Observations from practice — mainly on the gap between FHIR go-live and sustainable data integrity. Condensations from mandates, mirrored against formal methodology.
What FHIR Validation Does NOT Check.
A FHIR bundle passes validation. On four loss patterns that can survive a formally valid transfer.
Read full article →Who Maintains the FHIR Profiles After Go-live?
Go-live is a project, operations is a discipline. On maintaining profiles, terminologies and mappings — and how to tell whether governance is alive.
Read full article →What Your FHIR Profile Does Not Say.
A FHIR profile is a contract between source and target system. What it fixes survives the mapping — what it leaves open is lost silently.
Read full article →Ready for a conversation?
Whether SILD Pilot, FHIR profiling, platform migration or a sparring session on your data architecture — we respond within 48 hours.
Initial call (30 min)
Free of charge and without obligation. We understand your use case, check fit and honestly say whether we are the right partner.
SILD Pilot or PoC
SILD Pilot: a defined transfer, review rules agreed in writing, binding fixed price once the scope has been jointly defined. Or a PoC for a specific use case — each with a clear go/no-go criterion before any next step.
Engagement or ongoing advisory (3–12 months)
Productive implementation with real data flow. Or: ongoing support for defined processes as a reliable extension of your team.