Skip to content

Getting Started

Use this page when you want the first complete SDK workflow, not only a demo.

By the end, you will have installed the SDK, created a project, run a baseline simulation, certified the output, and reviewed the resulting bundle.

Read before: Quickstart if you have not run iints demo yet.

Read next: Workflow Hub, MDMP Quickstart, or Hardware Hub, depending on your goal.

The Workflow In One View

  1. run a simulation
  2. certify the output data
  3. review the run report

If you mainly need help choosing folders, install extras, or repository vs package usage, read Installation first.

1) Create and Activate a Virtual Environment

python3 -m venv .venv
source .venv/bin/activate
python -m pip install -U pip

All commands below assume this .venv is active.

2) Install The SDK

python -m pip install -U "iints-sdk-python35[full,mdmp]"

Optional extras:

pip install "iints-sdk-python35[research]"
pip install "iints-sdk-python35[nightscout]"
pip install "iints-sdk-python35[edge,mdmp]"

3) Verify The Environment

iints doctor --smoke-run

If this fails, fix the environment before starting longer runs.

4) Create A Project

iints quickstart --project-name iints_quickstart
cd iints_quickstart

The generated structure includes: - algorithms/ - scenarios/ - patients/ - contracts/ - data/demo/ - audit/ - results/

Important: - before iints quickstart, commands can be run from any folder - after iints quickstart, move into the generated project folder - repository helper scripts such as ./scripts/run_live_stage_demo.sh belong to the SDK repository, not the quickstart project - custom algorithms can be installed later with iints plugin install algorithms/my_algo.py

5) Run A Baseline Simulation

iints run --algo algorithms/example_algorithm.py \
  --patient-config-path patients/stable_patient.yaml \
  --scenario-path scenarios/clinic_safe_baseline.json \
  --duration 1440

6) Certify The Run Data

Use the generated results CSV plus the project contract:

iints data certify \
  contracts/clinical_mdmp_contract.yaml \
  results/<run_id>/results.csv \
  --output-json results/<run_id>/certification.json

Optional dashboard:

iints data certify-visualizer \
  results/<run_id>/certification.json \
  --output-html results/<run_id>/certification_dashboard.html

7) Review The Run With The Local AI Layer

iints ai report results/<run_id>

That gives the main SDK workflow in three commands:

  1. iints presets run
  2. iints data certify
  3. iints ai report

8) Inspect The Outputs

A typical run writes: - results.csv: time-series simulation output - clinical_report.pdf: summary report for review - audit/: safety and decision trail - run_manifest.json: file hashes for reproducibility - run_metadata.json: run configuration, environment details, SDK version, and data-format versions - certification.json: trust grade and dataset checks after iints data certify

9) Build A Study-Ready Bundle

iints study-ready --algo algorithms/example_algorithm.py --output-dir results/study_ready

This adds: - validation_report.json - sources_manifest.json - SUMMARY.md

10) Pick Your Next Workflow

The IINTS-AF SDK supports many advanced workflows that are outside the scope of a basic simulation, including:

  • Importing personal pump/CGM data (CareLink, Tidepool, Nightscout).
  • Running local AI Models (Ollama) to automatically audit and explain simulation runs.
  • Generating Poster Summaries for academic conferences.
  • Deploying to Edge devices like the Raspberry Pi or Arduino UNO Q for live hardware testing.

To keep this Getting Started guide clean, all these advanced flows have been moved to their respective guides. Please refer to the Workflow Hub, Raspberry Pi Digital Patient, or the AI Assistant guide for full commands on these topics.

Where To Go Next

If you finished this page and want to... Continue with
turn one run into a formal benchmark Scientific Workflow
aggregate several runs Study Analysis
certify datasets and outputs MDMP Quickstart
use local AI reporting AI Assistant
deploy to Raspberry Pi Raspberry Pi Digital Patient

More references:

Safety Scope

  • Research use only.
  • Not a medical device.
  • No clinical dosing advice.