Skip to Content

From AI Model to Clinical Application: Privacy-by-Design Prototyping in Python

TEF-Health Service
Consulting Virtual
Pricing/Discount Options: Call #2

Service Description

Overview:

Many health AI projects stall between a working model and a tool that clinicians can actually use. Building a secure client-server application usually requires a separate web team, a second codebase in JavaScript, and a rewrite before going to production.

How can the service help you?

CETIC builds interactive, multi-user applications around your AI models with Pyplet, its open-source (Apache 2.0) full-Python application framework. Python runs on both sides: a real Python interpreter in the browser (WebAssembly, with NumPy, SciPy and pandas) and a Python server, connected by a permanent bidirectional channel. This enables:

  • Privacy by design: sensitive data (e.g. medical images) can be processed directly on the user's workstation, while the server handles accounts, persistence and collaboration. This supports GDPR-compliant pilots.
  • Built-in governance: OAuth2/OIDC authentication (Google, Microsoft Entra), passwordless magic links, access control rules on pages and connections, user sessions.
  • Zero installation: users only need a browser and a URL; the server can run on-premise, inside the hospital network.
  • No throw-away prototype: the same code evolves from first demo to a hardened, tested tool. Your data scientists can keep maintaining it, since it is all Python.
How will the service be delivered?
    1. Needs and workflow analysis with end users (clinicians, technicians) and your team.
    1. Rapid prototype around your model, typically within 3 to 6 weeks.
    1. Pilot deployment with real users, on your infrastructure or on-premise at the clinical site, and iteration on feedback.
    1. Hardening: automated tests, packaging, documentation and handover of the full source code.

Typical applications: AI-assisted review interfaces (the clinician sees and corrects the AI output), data collection tools for clinical studies, interactive decision-support prototypes, on-device pre-processing of sensitive data before upload.

Reference: with Pyplet, CETIC built in a few weeks an imaging analysis tool now used in production by a research centre: data opens 40× faster and expert labelling is 10× faster. An internal AI document assistant was built on the same base in 3 weeks.

Keywords: human oversight GDPR AI deployment clinical application rapid prototyping Python web application privacy by design on-premise decision support open source
Service categories: Model & Algorithm (Development, Optimization & Evaluation, etc.) Infrastructure (compute, storage, network, HPC, etc.) Software (Cybersecurity and privacy-protecting enhancements, development, maintenance, deployment, administration, etc.)
Provider Logo

Provider & Contact

Provider Country Belgium
Organisation Website https://www.cetic.be/
Billing: per person-day
Full Price 1144 EUR
Reduced Price 858 EUR
Pricing Detail

Reduced: The reduced price value is indicative for a 10 day contract and can vary depenting on the total amount of effort and capacity of the service provider. Total effort depends on the applicant needs. The final price will be determined in the contract between the service provider and the applicant.

Operational Details

Service Inputs Trained AI model (or model API), sample data, description of the target clinical workflow, access to end users for feedback, hosting constraints.
Service Outputs Web-based multi-user application around the model, pilot deployment (on-premise or client infrastructure), automated tests, documentation, full source code.