IINTS-AF SDK Documentation¶
IINTS-AF is a research and education SDK for insulin-algorithm simulation, glucose-data quality review, local AI experiments, and bench-only hardware workflows.
Scope
IINTS-AF is not a medical device, does not provide clinical dosing advice, and is not intended for real insulin delivery. Use it for simulation, teaching, benchmarking, documentation, and controlled bench research.
Start Here¶
| Need | Page | Command |
|---|---|---|
| Install and verify the SDK | Quickstart | iints doctor --smoke-run |
| Choose the right workflow | Choose Your Path | iints guide |
| Look up practical commands | Command Cheatsheet | iints --help |
| Prepare a live demonstration | Booth Demo & Presentation | iints demo eucys |
| Keep the install current | Updating The SDK | iints update |
| Explain the safety boundary | Project Boundaries | iints safety-visualize |
| Understand sources and assumptions | Complete Source Library | iints sources |
| Work with hardware | Hardware Hub | iints edge doctor |
What The SDK Covers¶
| Area | What it does | Main pages |
|---|---|---|
| Simulation | Runs virtual-patient scenarios with algorithms, safety supervision, and reproducible outputs | Getting Started, Scientific Workflow |
| Data quality | Imports CGM/pump data, checks realism, and creates MDMP-style certification artifacts | MDMP Quickstart, Real-Data Realism Gate |
| Research AI | Tracks local AI setup, Mistral model migration, and public/request-gated diabetes datasets for research | AI Assistant, Mistral Model Migration, Local AI Research |
| Reports | Generates run reports, evidence bundles, posters, and AGP-style research glucose summaries | Research Evidence Bundle, Command Reference |
| Edge hardware | Supports Raspberry Pi, Jetson endurance runs, bench-only Pico/UNO workflows, and FPGA safety-core experiments | Hardware Hub, Jetson Endurance Mode, FPGA Mode |
| Development | Documents architecture, API symbols, contribution checks, and release maintenance | Developer Portal, API Reference, Maintainer Guide |
Core Workflow¶
- Configure a patient, scenario, algorithm, seed, and safety settings.
- Run a simulation or long study and preserve the output bundle.
- Validate results with realism, safety, and reproducibility checks.
- Package evidence through reports, manifests, plots, and citations.
First Commands¶
python3 -m venv .venv
source .venv/bin/activate
python -m pip install -U pip
python -m pip install -U "iints-sdk-python35[full,mdmp,research,edge]"
iints doctor --smoke-run
iints update --dry-run
iints demo eucys --output-dir results/live_demo
For source-install testing from the latest GitHub version:
python -m pip install -U "iints-sdk-python35[full,mdmp,research,edge] @ git+https://github.com/python35/IINTS-SDK.git"
What To Read Next¶
| If you are... | Read next |
|---|---|
| A first-time user | Quickstart then Getting Started |
| Preparing for a jury or booth demo | Booth Demo & Presentation then Command Cheatsheet |
| Training local AI models | Diabetes Research Datasets then Local AI Research |
| Reviewing evidence | Complete Source Library then Research Evidence Bundle |
| Building hardware demos | Hardware Hub then the board-specific guide |
| Maintaining the SDK | Developer Portal then Contribute Safely |
Project Boundary¶
IINTS-AF is useful for asking research questions such as whether a simulation is reproducible, whether a glucose trace is plausible, whether a controller behaves safely in a virtual patient, and whether a demo can be explained transparently.
It is not proof that an insulin algorithm is clinically safe. Any real-world medical use would require clinical validation, regulatory review, cybersecurity review, hardware verification, and qualified medical oversight outside the scope of this SDK.