The DIY Ecosystem
Before commercial AID systems existed, the #WeAreNotWaiting movement reverse-engineered insulin pumps and CGMs to create their own closed loops. These systems run on smartphones and communicate with pumps via Bluetooth or radio bridges (like RileyLink).
| System | OS | Algorithm Base | Key Feature |
|---|---|---|---|
| AndroidAPS (AAPS) | Android | OpenAPS / Oref1 | Super Micro-Boluses (SMB) |
| Loop (iOS) | iOS | Loop / Tidepool | Native Apple Health Integration |
| FreeAPS X | iOS | OpenAPS / Oref1 | AAPS logic on iPhone |
| iAPS | iOS | OpenAPS (Swift) | Dynamic ISF/CR adjusting |
Algorithm: Oref1 and SMB
The core logic of AAPS and FreeAPS X is Oref1. Unlike commercial systems that rely primarily on modifying basal rates, Oref1 utilizes Super Micro-Boluses (SMB). If the algorithm predicts glucose will rise above target, it delivers a portion of the required correction immediately as a bolus, bypassing the slow drip of basal insulin.
Furthermore, DIY systems support "Unannounced Meals" (UAM). The algorithm detects rapid deviations in glucose from carbohydrate absorption and delivers aggressive SMBs without the user inputting carbs, attempting to flatten the curve autonomously.
Hardware Requirements
DIY systems require specific hardware combinations, often relying on older pumps with known security exploits, or newer pumps with official but localized Bluetooth APIs (e.g., Dana-i, Omnipod DASH).
- Omnipod DASH: Currently the most popular choice for DIY as it communicates natively via Bluetooth without external bridges.
- Dana-i / Dana RS: Official Korean pumps designed for open-source looping.
- Older Medtronic (e.g., 722, 523): Require an OrangeLink/RileyLink to translate BLE to 900MHz radio.
Dynamic Settings (Autosense)
DIY systems evaluate your insulin sensitivity continuously. During exercise, illness, or hormonal shifts in pregnancy, "Autosense" scales your programmed ISF and Basal profile up or down, acting as a real-time calculator.
FAQ
- Is this legal/safe?
- Building the app from source code is legal. However, it is entirely unapproved by the FDA/EMA. You assume all risk. There is no customer support, only community documentation. Thorough troubleshooting skills are mandatory.
- Why doesn't everyone use this?
- The barrier to entry is high. You must compile an Xcode or Android Studio project on a computer, manage Apple Developer certificates, and deeply understand data parameters to tune the system safely.