Diabeloop DBLG2: FDA-Cleared and CE-Marked Advanced AID Algorithm Designed to manage unexpected glycemic excursions

Editor’s Note: This release was updated on March 3rd, 2026 to more precisely reflect the FDA-cleared Intended Use and the specific design capabilities of the DBLG2 algorithm.

Grenoble, France — March 3rd, 2026 — Diabeloop, a pioneer in therapeutic artificial intelligence, is highlighting the advanced capabilities of its DBLG2 algorithm, recently cleared by the U.S. Food and Drug Administration (FDA) and CE Marked. The DBLG2 algorithm is engineered to provide a high degree of automated support and flexibility by continuously managing glucose trends. The algorithm is designed to automatically deliver correction boluses in response to glucose values exceeding a predefined threshold. This automated correction mechanism serves as a critical support feature for managing unexpected glycemic excursions.

Adaptive Automated Correction for Hyperglycemia

As an Interoperable Automated Glycemic Controller (iAGC) with both FDA clearance (Class II) and CE marking (class III), DBLG2 is engineered to provide a high degree of automated flexibility. The DBLG2 algorithm is also designed to automatically manage glycemic excursions and deliver adapted correction boluses.

Patients should announce their meals for optimal glycemic control (as written in the labelling). However, in real-world observations from the commercially available version that Diabeloop introduced in Europe in early 2025, data showed that users occasionally may not announce all meals. From a pool of 800,000 patient-days, it was observed that meals were not manually declared approximately 9% of the time. In these instances, the DBLG2 algorithm’s ability to automatically adjust insulin was evaluated to understand its impact on patient safety and glycemic control.

Clinical Performance and Safety Observations (data for DBLG1 in Europe)

These analyses are provided as evidence of algorithm behavior in certain situations, and should not be mis-interpreted as a recommendation to not declare meals: users should follow labeling and declare their meals in order to maintain optimal glycemic control. 

Multiple studies were conducted with DBLG1 in real-world use (in the EU). In these studies, the average loss of Time-in-Range varied between 3 and 6 percentage points when no meals were announced during the day compared with full meal announcement. Overall hypoglycemia and metabolic events levels were not affected. Significant individual variability was observed, meaning these results are averages only, with degraded glycemic control in certain individuals:

• Study SP13 (NCT04725591) – In DBLG1 real-world use (EU), this Randomized controlled clinical trial, conducted on 49 adolescents with Type 1 Diabetes, analyzed the safety and performance of DBLG1 when no meals are announced in this population. Specifically, the Time-in-Range (TIR) remained statistically non-inferior to the standard use where all meals are announced (-5.89 percentage points), and almost 14 percentage points higher compared to usual treatment.
• Study Obs32 (Real-world evidence) –In DBLG1 real-world use (EU), this large real-world analysis, based on 1,212 adults treated with DBLG1 corresponding to 118,598 patient-days,  the observed  median Time-in-Range (TIR) is  almost 70% during days with no meal announced. This was 3 percentage points lower compared to the median for all patient-days, while the time spent in hypoglycemia (below 54 mg/dL) remained very low and well below international recommendations.
• Study Obs36 (Real-world evidence) –In DBLG1 real-world use (EU), this large real-world analysis, based on 2,335 adults treated with DBLG1, a subgroup analysis based on patients’ meal declaration habits showed that the patients declaring less than one meal per day on average reached a mean Time-in-Range of 67% (compared to 72.5% for patients declaring three or more meals per day), with time spent in hypoglycemia maintained well below international recommendations. In addition, no severe metabolic events were observed in this specific patient subgroup.

Advancing Innovation in Diabetes Management

Diabeloop remains committed to research that prioritizes patient well-being and clinical safety. Recent and ongoing innovation workstreams are being part of Diabeloop’s long-term vision to further reduce the manual interventions required for effective glycemic control. 

“Reducing the impact of glycemic variability has been a longstanding goal for automated insulin delivery systems,” said Erik Huneker, co-founder and Chief Science Officer. “Diabeloop software’s ability to adapt in real-time and address unexpected glycemic excursions offers an important layer of automated support in managing diabetes.”

“The next frontier for Diabeloop’s research lies in physical activity and movement detection, another key component in glucose variability that remains to be seamlessly integrated into autonomous insulin management. This advancement reinforces Diabeloop’s commitment to innovation centered around patient freedom and well-being. Our teams are already advancing the next generation of sensor-driven features to further personalize and automate treatment in response to lifestyle adaptation.” added Erik Huneker 

About Diabeloop

Founded in 2015, Diabeloop develops artificial intelligence-based solutions to help people living with diabetes automate and personalize their treatment. By connecting continuous glucose monitors (CGMs) and insulin pumps through proprietary self-learning algorithms, Diabeloop enables users to delegate many therapeutic decisions and live with fewer daily interruptions. 

DBLG1, Diabeloop’s first marketed system in Europe, has been available since 2021 and is interoperable with Kaleido® (ViCentra), Dana-i® (Sooil), MEDISAFE WITH® (Terumo), and Accu-Chek® Insight (Roche) insulin pumps. 

DBLG2 received CE marking under the EU MDR in August 2025 and is currently in its pre-launch phase in Europe, a full launch is planned for early 2026, including Kaleido and Dana-i insulin pumps. 

In December 2025, DBLG2 was granted FDA clearance as a Class II Interoperable Automated Glycemic Controller (iAGC). The indication covers people with Type 1 diabetes aged 12 years and older. 

For more information, visit Diabeloop.com or send an email at contact@diabeloop.com