Aims: Open-source automated insulin delivery (AID) predated the availability of commercial systems and is used by thousands with type 1 diabetes (T1D) despite no regulatory approval. We examined efficacy and safety of an open-source AID system.
Methods: A 24-week, multi-center randomised (1:1) controlled trial in children (7–15 years) and adults (16–70 years) with T1D, comparing open-source AID (OpenAPS algorithm within a modified version of AndroidAPS in a smartphone, pre-production DANA-i™ insulin pump, and Dexcom G6® CGM), to sensor augmented pump therapy (SAPT). The primary outcome was percent time in target sensor glucose range (TIR; 3.9–10.0 mmol/L), between AID and SAPT during days 155-168 (final 2 weeks of the study).
Results: Ninety-seven participants (48 children, 49 adults) were randomised (44 to open-source AID and 53 SAPT). The mean adjusted difference in TIR between AID and SAPT at end of study was 14% (95% confidence interval (CI), 9.2 to 18.8; P<0.001), with no treatment effect by age interaction (p=0.56). AID users spent 3 hours 21 minutes (95% CI, 2h 12m to 4h 30m) more in target range per day. In the AID arm, mean TIR (±SD) increased from 61.2±12.3% to 71.2±12.1%, and decreased from 57.7±14.3% to 54±16% in the SAPT arm. More participants achieved TIR >70% using AID (60% vs. 15%). No severe hypoglycemia or diabetic ketoacidosis occurred in either arm. Two participants withdrew from AID due to connectivity issues.
Conclusion: Open-source AID using the OpenAPS algorithm within a modified version of AndroidAPS, a widely used open-source AID solution, is efficacious and safe.