Klinische Entscheidungsunterstützung, entwickelt für Pflegekräfte.
ARIA is a clinical decision support system (CDSS) built from the ground up for the people who carry the load in acute and intensive care: nursing staff. The platform continuously consolidates vital signs, lab results, and nursing context into explainable risk scores and prioritized recommendations — surfacing clinical deterioration earlier, reducing documentation burden, and keeping clinical authority where it belongs: with the human at the bedside.
Existing CDSS tools (Epic Sepsis Watch, Bayesian Health, Sana AI) are designed around physician workflows and reach nursing acceptance rates below 40 %. ARIA closes that structural gap. Three explicit goals: early warning of clinical deterioration with a measurable lead time (median 4 hours ahead of standard scores), 15–25 % reduction in documentation-driven nursing workload, and per-datapoint explainability — a precondition for both clinical adoption and MDR certification.
| Feature | What it does |
|---|---|
| Risk-prioritized ward dashboard | All patients of the shift, sorted by clinical urgency |
| Live vital monitoring | Trend curves from wearables and EHR, ~30 s refresh |
| Explainable AI recommendation | Recommendation plus top-3 weighted factors (SHAP) |
| SBAR handover assistant | Auto-generated shift handover from longitudinal data |
| Anomaly detection | Subtle deviations from a patient's personal baseline |
| Audit trail | Append-only log with hash-chain for full traceability |
| Layer | Stack | Status |
|---|---|---|
| Frontend | Next.js 16 (App Router), React 19, TypeScript strict, Tailwind v4 | In MVP |
| Backend (core) | Node.js + Fastify, OpenAPI-generated clients | Planned |
| Backend (ML) | Python + FastAPI, ONNX Runtime, MLflow | Planned |
| Database | PostgreSQL (Neon EU), Redis, Object Storage (Hetzner) | Planned |
| ML stack | XGBoost (risk), Temporal Fusion Transformer (forecast), German Medical BERT (NLP) | Planned |
| Interop | HAPI FHIR R4 gateway, HL7 v2.5, MQTT for bedside sensors | Planned |
| Hosting | Vercel (frontend, EU regions) + Hetzner Cloud Falkenstein (backend, GPU) | Planned |
| Auth | Auth.js + Microsoft Entra ID (clinic AD integration) | Planned |
Full rationale and architectural decisions: see docs/ARCHITECTURE.md.
git clone https://github.com/Orouji-Mo7/aria-platform.git
cd aria-platform
npm install
npm run devThe dev server starts on http://localhost:3000.
Production build:
npm run build
npm run startMVP — frontend only. This repository currently ships the dashboard shell with synthetic data only. There is no real patient context, no EHR integration, no live ML pipeline, and no certified medical-device functionality. ARIA is not approved as a medical product and must not be used for clinical decision-making.
Roadmap, certification path (CE Class IIa under MDR), data architecture, and security model are documented in docs/ARCHITECTURE.md.
- docs/ARCHITECTURE.md — full technical architecture, product specification, business model, and roadmap (German)
Mohammad Orouji
M.Sc. Digital Health Management — MSH Medical School Hamburg
Background: Pflegefachkraft (20+ years of clinical practice) · Fachinformatiker für Anwendungsentwicklung (IHK 2025)
ARIA is built on the conviction that clinical software for nurses should be designed by people who have stood at the bedside.
This project is released under a custom proprietary license — see LICENSE for full terms.
Code is publicly visible for transparency and academic review. Commercial use, redistribution, modification, or derivative works require prior written permission from the copyright holder.
Copyright © 2026 Mohammad Orouji. All rights reserved.
For licensing inquiries, please reach out via my GitHub profile or LinkedIn.