Loop is an open-source automated insulin delivery app for iPhone and Apple Watch. It reads glucose from a continuous glucose monitor (CGM), forecasts glucose from insulin, carbs and recent trends, and adjusts delivery on a compatible insulin pump. It is a stone resting on the boulders of work done by many others.
Loop is built on LoopKit, a set of frameworks for data storage, retrieval and calculation. Pump and CGM support comes from separate plugins, and the app is built from LoopWorkspace.
Please understand that this project:
- Is highly experimental
- Is not approved for therapy
Screenshots are from the iOS Simulator with simulated CGM and pump data.
- Automated insulin delivery. Loop forecasts glucose several hours ahead and adjusts delivery with each new glucose reading, using either temporary basal rates or automatic boluses.
- Meal and correction boluses with a recommended amount and a forecast of where glucose is heading. Carb entries can include food type, absorption time and saved favorite foods.
- Pre-meal targets and custom presets that temporarily adjust overall insulin needs and the correction range, for exercise, illness or other situations. Presets can run for a set time or indefinitely, and can be scheduled to start later.
- Live Activity on the Lock Screen, in the Dynamic Island and on CarPlay.
- Apple Watch app with glucose and forecast, carb entry, bolus, pre-meal targets and presets.
- Alerts for glucose and device conditions, with critical alerts that sound even when the phone is silenced, and an option to temporarily mute alerts.
- Algorithm experiments: optional Glucose Based Partial Application and Integral Retrospective Correction.
- Data sharing with Apple Health, and upload to Nightscout and Tidepool through optional services.
- Pumps: Omnipod 5, Omnipod DASH and Omnipod Classic (Eros, with RileyLink); older Medtronic pumps (with RileyLink); Medtrum Nano (200U and 300U).
- CGMs: Dexcom G5, G6, ONE, G7 and ONE+; FreeStyle Libre 1 and 2 (through MiaoMiao or Bubble transmitters, or EU Libre 2 directly); Eversense E3 and 365; and readings from Dexcom Share or Nightscout.
See Loop Docs for the exact models and requirements.
Please visit the Loop Docs for installation, algorithm, and other details. Loop is built from LoopWorkspace, in a web browser with GitHub Actions or on a Mac with Xcode.
For FAQs, how to contribute to open source aspects of Loop and other tips, refer to the Wiki.
Questions and discussion are welcome on Loop Zulip.
Please read the LICENSE and CODE_OF_CONDUCT