iAPS

iAPS is an open-source automated insulin delivery (AID) app for Apple iPhones that runs the OpenAPS "oref" dosing algorithm on-device, automatically adjusting insulin from continuous glucose monitor (CGM) readings.¹ It grew out of Ivan Valkou's FreeAPS X — an iOS port of OpenAPS — which after thousands of community commits was rebranded as iAPS under the "Artificial-Pancreas" organization.¹

Platform. iAPS is iOS-only: it needs an iPhone 8 or newer running at least iOS 17, with no Android version.¹ (Its closest Android counterpart is AndroidAPS, which runs the same oref algorithm.)

Algorithm. iAPS is built on the OpenAPS Reference (Master 0.7.1) algorithm and also borrows frameworks from the Loop community.¹ oref every five minutes predicts glucose and sets a temporary basal or micro-bolus. On top of the standard model, iAPS exposes autosens (which reviews the last 8 and 24 hours to nudge basal, target and insulin sensitivity), dynamic ISF and dynamic carb ratio (which scale sensitivity with glucose and total daily dose), and autotune.² An optional JavaScript "middleware" layer lets advanced users inject custom dosing logic.¹

Supported pumps & CGMs. Pumps include Omnipod EROS and DASH, Dana-i and DanaRS (firmware 3), Medtrum TouchCare Nano, and several Medtronic models.¹ CGMs include Dexcom G5, G6, ONE, ONE+ and G7, European Libre 2 and 2 Plus, Medtronic Enlite, and Nightscout as a CGM source.¹

Customizability. Because every parameter is open and inspectable, iAPS is among the most tunable loops available — the defining trade-off of the DIY approach, which buys flexibility commercial systems don't offer at the cost of self-reliance.³

Setup & legal status. iAPS is explicitly described by its developers as "highly experimental and rapidly evolving" and "not CE or FDA approved for therapy."¹ There is no commercial product to buy: you build the app yourself in Xcode or via GitHub Actions, and some pump connections require a paid Apple developer account.¹ Using it is a self-responsibility, do-it-yourself undertaking — the user, not a manufacturer, owns the outcome.

Clinical evidence. No randomized trial has tested the iAPS app by name. The strongest evidence for its class is the CREATE randomized controlled trial, which paired the same oref/OpenAPS algorithm (via AndroidAPS) with a pump and Dexcom G6: time-in-range rose to 71% with AID versus a sensor-augmented pump, an adjusted 14-percentage-point gain (about 3 hours 21 minutes more per day in range) with no severe hypoglycemia or ketoacidosis.⁴ The 48-week continuation phase confirmed a sustained ~12-point time-in-range advantage and a 0.5% HbA1c reduction across different pumps.⁵ A systematic review of real-world data likewise found open-source AID systems (OpenAPS, AndroidAPS, Loop) safe and effective.⁶ These results transfer to iAPS to the extent it shares the oref algorithm, but are not direct evidence for the iAPS app.

What's coming. iAPS remains actively maintained — releases continued into 2026 (version 8.2.1 in May 2026), with recent work on the Auto-ISF feature and an Apple Watch interface.⁷ However, after iAPS version 3.0.0, disagreements over development process, open-source principles and peer review led part of the community to fork the project into Trio, now developed under the Nightscout Foundation.⁸⁹ Trio has drawn the larger share of active community momentum, and the ecosystem's documented migration path now points iAPS users toward Trio.¹⁰ iAPS is still installable and usable today, but its forward trajectory is increasingly defined by that successor project rather than by major new feature releases of its own.

References

1. Artificial-Pancreas. iAPS — README (project overview, algorithm, compatible pumps/CGMs, iPhone/iOS requirements, experimental/not-approved status). GitHub (accessed 2026). https://github.com/Artificial-Pancreas/iAPS ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

2. iAPS Project. Autosens, Dynamic ISF/ICR and Autotune. iAPS Documentation (Read the Docs) (accessed 2026). https://iaps.readthedocs.io/en/dev/settings/configuration/concepts/autosens-dynamic.html ↩

3. Burnside MJ, Lewis DM, et al. CREATE trial protocol — note on open-source algorithms' capacity for individualisation versus commercial AID. J Diabetes Metab Disord (2020). https://doi.org/10.1007/s40200-020-00547-8 ↩

4. Burnside MJ, Lewis DM, Crocket HR, et al. Open-Source Automated Insulin Delivery in Type 1 Diabetes (CREATE randomized controlled trial). N Engl J Med (2022);387(10):869–881. https://doi.org/10.1056/NEJMoa2203913 ↩

5. Burnside MJ, Lewis DM, Crocket HR, et al. Extended Use of an Open-Source AID System: the 24-Week Continuation Phase Following the CREATE RCT. Diabetes Technol Ther (2023);25(4):250–259. https://doi.org/10.1089/dia.2022.0484 ↩

6. Knoll C, Peacock S, Wäldchen M, et al. Real-world evidence on clinical outcomes of people with type 1 diabetes using open-source and commercial automated insulin dosing systems: a systematic review. Diabet Med (2021);39(5):e14741. https://doi.org/10.1111/dme.14741 ↩

7. Artificial-Pancreas. iAPS Releases (v8.2.1, May 2026; v8.2.0; Auto-ISF and Watch updates). GitHub (accessed 2026). https://github.com/Artificial-Pancreas/iAPS/releases ↩

8. Trio Project. FAQ — history of the fork from FreeAPS X to iAPS to Trio, and the post-3.0.0 split over development/open-source/peer-review. TrioDocs (accessed 2026). https://triodocs.org/help/faq/ ↩

9. Nightscout Foundation. Trio — automated insulin delivery system for iOS based on the OpenAPS algorithm (repository, created 2024). GitHub (accessed 2026). https://github.com/nightscout/Trio ↩

10. Trio Project. Migrating from iAPS. TrioDocs (accessed 2026). https://triodocs.org/configuration/migration/iaps-migration/ ↩