Project

InsuLearner

My son was diagnosed with Type 1 Diabetes in 2017. This is the tool I built to estimate his insulin pump settings from his own data, now free and open-source for anyone who needs it.

InsuLearner data feature visualization from project workflow.

I developed this approach while working as a data scientist at Tidepool, building risk simulators for their FDA-cleared insulin dosing system. The same methodology is now packaged and free for anyone to use. It estimates Basal Rate, Carbohydrate Ratio (CIR), and Insulin Sensitivity Factor (ISF) from existing pump and CGM data. It works for insulin pumps, closed-loop systems, and manual daily injections. T1D families from around the world have used it.

Insulearner Web

Free web app. No coding required. Start here if you want to run the algorithm without installing anything.

Open Insulearner Web

InsuLearner on GitHub

The open-source Python package. Technical source of truth. Current version: insulearner==0.2.0. Tidepool and Nightscout supported.

Open on GitHub

Writing

These tell the full story, in order.

  1. How I Calculate My Son's Insulin Pump Settings with Machine Learning
  2. InsuLearner Software for Computing Insulin Pump Settings
  3. InsuLearner + Insulearner Web: Nightscout Update

Medical Disclaimer

I am not a doctor and this is not medical advice. Check with your doctor before making any changes to insulin pump settings.