Summary: In 28 adults with type 1 diabetes who had never used automated insulin delivery, a tubeless open-source hybrid system (AndroidAPS) was compared with sensor-augmented pump therapy in a 26-week crossover trial. Time in range over the final two weeks of each phase was 75.6 percent with the open-source system versus 60.4 percent with the pump (p < 0.01), with no increase in time below range and no severe hypoglycaemia or serious adverse events.
PICO Summary
| Element | Detail |
|---|---|
| Population | 28 adults with type 1 diabetes (mean age 33.6 years, mean diabetes duration 12.6 years, HbA1c 7 to 11 percent), AID-naive; single-centre, open-label randomised crossover trial in China. |
| Intervention | Tubeless open-source hybrid automated insulin delivery (Android version, AAPS) used at home for 3 months (n = 28 contributing to each phase via crossover). |
| Comparison | Sensor-augmented insulin pump (SAP) therapy for 3 months (same 28 participants, crossed over). |
| Outcome | Primary endpoint, percentage time in range (70 to 180 mg/dL) during the final two weeks of each phase: 75.6 plus or minus 10.3 percent (AAPS) versus 60.4 plus or minus 15.1 percent (SAP), p < 0.01. Nocturnal time in range exceeded daytime (80.9 plus or minus 13.6 percent versus 73.8 plus or minus 10.2 percent, p < 0.01). No significant between-phase difference in time below range; no severe hypoglycaemia or serious adverse events in either phase. Confidence intervals, absolute risk reduction and number needed to treat were not reported. |
Tubeless open-source AID vs SAP in type 1 diabetes
RCT crossover · type 1 diabetes · 26 weeks
A tubeless open-source hybrid loop raised time in range by about 15 percentage points over sensor-augmented pump therapy, with no increase in hypoglycaemia. Small single-centre crossover trial in AID-naive adults.
Expert Commentary
This crossover trial reports a clinically meaningful 15 percentage point gain in time in range with a tubeless open-source automated insulin delivery loop over sensor-augmented pump therapy, achieved without any rise in time below range or any severe hypoglycaemia. The within-patient crossover design strengthens the comparison by controlling for individual variation, and the magnitude of nocturnal benefit is consistent with what algorithm-driven overnight basal modulation should deliver. The result is encouraging but must be read with restraint. The sample is small at 28 participants from a single centre, the trial was open-label so participants and clinicians knew their assigned therapy, and the population was AID-naive, which can inflate the apparent benefit of any first automated system. The system is an off-label, do-it-yourself loop rather than a regulator-approved device, and durability beyond 26 weeks, generalisability outside this Chinese cohort, and harder endpoints such as HbA1c and severe-event rates remain to be confirmed in larger, longer studies. Funding was non-commercial, which reduces sponsorship concern. Can I use this with my patients? Not yet as a recommended standard, but it offers reassuring signal for a motivated, technically capable patient already considering an open-source loop, who should be counselled that the approach is unapproved and self-managed. A larger blinded-endpoint trial with HbA1c outcomes is needed before broader adoption.
References
Lei M, Ling P, Lin B, Lv J, Liu Z, Zhou Y, et al. Efficacy and safety of a tubeless open-source hybrid automated insulin delivery use at home among adults with type 1 diabetes mellitus: Results from a 26-week, free-living, randomized crossover trial. Diabetes Obes Metab. 2025;27(9):4699-4708. doi:10.1111/dom.16505