Automated Insulin Delivery Systems Research Articles (Page 1)

This study evaluated the glycemic responses to a graded exercise test (GXT) performed by 24 adolescents with type 1 diabetes (T1D) using automated insulin delivery (AID) systems. Each participant partook in a GXT on a bicycle ergometer until volitional exhaustion. Plasma glucose and lactate levels were measured during the GXT, whereas sensor glucose was monitored in the hours thereafter. Plasma glucose levels were stable throughout the GXT (overall change of -0.26 mmol/L [-5 mg/dL], P = 0.593), with no hypoglycemic events. Sensor glucose levels also remained stable and within the recommended glucose target ranges after the GXT for the remaining day and night, with only a few episodes of mild hypoglycemia. This study highlights the glycemic safety of performing GXT for adolescents utilizing AID systems.

OBJECTIVE Automated insulin delivery (AID) systems improve glycemic outcomes in youth with type 1 diabetes and are now the recommended mode of insulin delivery. Previous studies highlighted racial disparities in the use of continuous glucose monitoring and insulin pump therapy. The purpose of this study was to evaluate the use of AID systems and A1C outcomes in youth with type 1 diabetes by race. RESEARCH DESIGN AND METHODS This was a single-center cross-sectional study. We included youth and young adults with type 1 diabetes aged 2–21 years who had at least two clinic visits between December 2022 and December 2023. Demographics, diabetes device use, and A1C data were gathered from chart review, based on the latest office visit records available in the electronic medical record system for 2023. RESULTS Out of 668 youth aged 3–20 years with type 1 diabetes, 435 (65%) were AID users. The prevalence of AID use was 70% (341 of 483) in White youth compared with 47% (60 of 129) among Black youth and 62% (34 of 55) among youth of other racial groups (P <0.001). Black youth using AID achieved significantly lower A1C levels (median 8%, interquartile range [IQR] 7.5–8.8%) compared with Black youth who did not use an AID system (median 9.6%, IQR 8.1–11.6%, (P <0.001). CONCLUSION These findings support the persistence of racial disparities in diabetes technology utilization.

Diabetic retinopathy remission in patients using an automated insulin delivery system: A prospective controlled study.

Effect of daily insulin delivery variability on glycemic outcomes in patients with type 1 diabetes mellitus using an automated insulin delivery system.

Human factor in the use of automated insulin delivery systems: Real-world effects in children with type 1 diabetes after 2years of use (AID-A study).

While automated insulin delivery (AIDs) systems have significantly improved glycemic control for individuals with type 1 diabetes (T1D), there remains a need for identifying and acting upon complex physiologic and behavioral patterns which consistently lead to hypo- and hyperglycemia. Prior methods have lacked the ability to automatically identify and extract patterns across mixed-type multidimensional data (eg, insulin, glucose, activity) without instilling bias from stipulations on time-lagged coupling, pattern length, or pre-defining patterns. We introduce a new pattern-detection technique-Block-based Recurrence Quantification Analysis (BlockRQA)-and preliminary results using BlockRQA in an AID on both in silico and in an outpatient feasibility study. We first introduce the BlockRQA algorithm, which extends Recurrence Quantification Analysis for use in categorical and continuous time-series data, while maintaining interpretable patterns in the domain of interest, in contrast to prior state-of-the-art approaches which require embeddings. Next, we demonstrate the feasibility of utilizing these patterns and BlockRQA with an existing AID system (BlockRQA+AID) to identify and dose for patterns leading to hyperglycemia in individuals with T1D. We demonstrate how BlockRQA+AID can improve glucose outcomes in patterns leading to hyperglycemia in silico. And we show real-world results using BlockRQA+AID to reduce hyperglycemic events (>250 mg/dL) via an interim safety analysis of a small outpatient pilot study. For all cases, we show BlockRQA efficiently identifies, aggregates, and scores behavioral patterns which can be targeted for clinical intervention. The BlockRQA is a powerful pattern recognition tool that may be used to identify glucose outcome patterns to guide AID dosing.

Automated insulin delivery (AID) systems continuously deliver insulin subcutaneously, reducing the burden of managing type 1 diabetes mellitus (T1D). However, there are limited data comparing different insulin delivery modalities, particularly regarding their impact on health-related quality of life (HRQoL). This study aimed to quantify the disutility associated with conventional insulin delivery modalities and utility gains associated with wearable, on-body, AID systems. Health state vignettes representing different insulin delivery modalities were developed based on interviews with people with T1D alongside published literature and validated by experts. Utility values were elicited via the time trade-off (TTO) method from the general population in the United Kingdom (UK) (n = 110). The lowest mean utility values were observed for tubed non-AID systems (0.727), while the highest mean utility value was observed for tubeless systems with AID (0.909). The use of tubeless systems rather than tubed systems was associated with a significant increase in utility between + 0.082 and + 0.086 (p < 0.005), and the use of AID was associated with a significant increase in utility of between +0.096 and +0.100 versus the corresponding alternatives (p < 0.0005). The use of a tubeless and AID system was associated with a significantly increased utility versus all other health states (p < 0.0001), indicating significantly higher HRQoL. This study elicited utility values for health states representing insulin delivery modalities in T1D. Results suggested that tubeless and AID systems are associated with higher health state utility in T1D, indicating that people with T1D using such systems may experience improved HRQoL.

The current study sought to evaluate a family-based programme designed for adolescents with type 1 diabetes and their parents and identify which families benefited most. A randomized controlled trial with a waitlist control with 157 parent-adolescent dyads collected data via online surveys and glycaemic measures at baseline and 3- and 6-month post-baseline. Regression models tested main and moderated effects on primary outcomes of percent time-in-range (% TIR) and diabetes health-related quality of life (HRQOL) and secondary outcomes of HbA1c and adolescent and parent diabetes distress. Beneficial intervention effects included improved parent report of supportive parenting (β = 0.151, p = 0.007, d = 0.31) and both parent (β = -0.177, p = 0.002, d = -0.36) and adolescent (β = -0.150, p = 0.024, d = -0.30) report of unsupportive parenting immediately after the intervention for the full sample. Improvements in HRQOL (β = 0.308, p = 0.002, d = 0.81) at 3-month post-baseline among racial and ethnic minoritized adolescents, HRQOL at 3-month (β = 0.261, p = 0.003, d = 0.94) and 6-month (β = 0.220, p = 0.005, d = 0.58) post-baseline among adolescents not using automated insulin delivery systems. The intervention also reduced parent diabetes distress among parents with high baseline distress (β = -0.200, p = 0.006, d = -0.54). While family dynamics improved for all, adolescents' quality of life improved among youth with fewer resources and from marginalized racial and ethnic backgrounds, whereas parents with higher distress benefited most. For adolescents, socioeconomic context may drive intervention response.

The use of automated insulin delivery systems (AID) is standard of care for people with type 1 diabetes. However, the limited capacity of insulin pump cartridges, which can hold 1.6-3.0mL or the equivalent of 160–300 units of U100 insulin, can be a barrier to AID use for individuals with high total daily insulin (TDI) requirements. With the rising prevalence of obesity, expansion of AID use to type 2 diabetes, and trends towards smaller cartridge volumes to decrease the size of devices, practical solutions to reduce barriers to AID use for those with high TDI requirements are needed. U200 concentrated rapid-acting insulin (U200) has a similar pharmacokinetic and pharmacodynamic profile to U100 insulin, provides the same dose of U100 insulin in half of the volume, and has been used off-label to facilitate AID use for those with high TDI needs. In this perspective piece we provide practical considerations for clinical implementation of U200 use in AID systems, including identification of candidates, unique considerations in filling pumps with U200 insulin, guidance on programming appropriate AID settings for the different algorithms, concepts to address in patient education, and recommendations for standardized documentation in the electronic health record.

Balancing continuity of care and liability concerns: One community pharmacy experience of a person with type 1 diabetes using a do-it-yourself automated insulin delivery system.

ABSTRACTBackgroundThis systematic review and meta‐analysis evaluated the efficacy and safety of the iLet bionic pancreas (iLet BP), a novel automated insulin delivery (AID) system, in managing type 1 diabetes. Unlike conventional AID systems, which require user input for insulin dosing, the iLet BP autonomously determines insulin delivery based solely on body weight. The study synthesized data from five randomized controlled trials (RCTs), comprising a total of 1130 patients, comparing iLet BP with standard care (SC).Outcomes AssessedPrimary outcomes included changes in HbA1c, mean glucose levels, and time in target glucose range (70–180 mg/dL), measured via continuous glucose monitoring (CGM). Secondary outcomes assessed adverse events and hypoglycaemia.Key FindingsResults demonstrated that the iLet BP significantly improved glycaemic control. The pooled analysis showed a standardised mean difference (SMD) in HbA1c of −0.50 [−0.63, −0.38] and in mean glucose levels of −0.36 [−0.50, −0.21] favouring iLet BP. Time in target glucose range was significantly higher with iLet BP (SMD: 0.58 [0.43, 0.73]). However, the odds of adverse events were notably higher in the iLet BP group (OR: 15.48 [8.07, 29.70]), while the risk of hypoglycaemia (OR: 2.22 [0.83, 5.94]) was not statistically significant.ConclusionIn conclusion, the iLet BP shows strong potential in improving glycaemic outcomes in patients with type 1 diabetes. However, concerns remain regarding its safety profile, particularly related to adverse events. Further large‐scale, high‐quality studies are needed to confirm its effectiveness and ensure broader clinical applicability.

Bridging the diabetes care gap in elderly: Lessons from automated insulin delivery systems

This study evaluated 1-year real-world changes in glycaemic management, parent-reported outcomes and safety with an automated insulin delivery (AID) system in young children with type 1 diabetes (T1D). Children aged 2-6 years whose parents agreed to initiate the Medtronic MiniMed™ 780G were enrolled at 15 centres between October 2022 and December 2023. Data were collected quarterly over 1 year during routine follow-up. Parent-reported outcomes were assessed using questionnaires (HAPPI-D[part of the HAPPI-D Protocol; Hvidøre, Adolescent, Parent, Professional, Instrument, Diabetes]and Hypoglycaemia Fear Survey [HFS]-Parent). The primary endpoint was the evolution of time in range (TIR, 70-180 mg/dL) from start to 12 months after start. Data are reported as mean ± SD or least-squares mean (95% confidence interval). A total of 149 children were included (mean age 4.2 ± 1.4 years; 56.4% girls). Mean T1D duration was 22.0 ± 13.0 months and 75.2% used an insulin pump before. After 1 year, TIR increased from 56.8% (54.4-59.2) to 66.6% (64.7-68.5) and haemoglobin A1c decreased from 7.6% (7.4-7.8) to 7.2% (7.1-7.4) (all p < 0.001). Time <70 mg/dL remained stable (5.0% [4.2-5.8] at start vs. 4.6% [3.9-5.3] at 12 months, p = 0.172). Parents reported less diabetes burden on the HAPPI-D (22.9 points [21.7-24.0] at start vs. 21.7 points [20.5-22.8] at 12 months, p = 0.001), while scores on the HFS-Parent did not change significantly. There were no hospitalisations for severe hypoglycaemic events and one for diabetic ketoacidosis due to infusion set occlusion. One-year use of an AID system in young children with T1D was safe and associated with improved glycaemic management and reduced parental burden, with limited impact on time in hypoglycaemia and related parental fear.

Factors Considered Important by Diabetes Providers Before Recommending Automated Insulin Delivery Systems: Observations From the T1D Exchange Quality Improvement Collaborative

Disclosure: S.J. Russell: Beta Bionics. R.R. Selagamsetty,: Beta Bionics. C.S. Balliro: Beta Bionics. E. Damiano: Beta Bionics.Background: The iLet bionic pancreas is an automated insulin delivery system that is initialized with only body weight. The only user input is meal announcements without carbohydrate counting. Meals and snacks are announced as a type (Breakfast, Lunch, or Dinner) and a relative size for that type (Usual for me, More, or Less). The iLet then autonomously determines the size of meal doses. All basal and correction insulin is determined autonomously by the iLet, which adapts continuously to changing insulin needs. The only input from the health care provider is the glucose target profile. Objective: The objective of this analysis was to assess the association between the number of meals announced by the user per day on one hand, and the frequency of user interaction with the iLet (determine by number of swipes to unlock the iLet) on the other, on continuous glucose monitoring (CGM) outcomes during real-world use of the iLet bionic pancreas by adults (≥18 years of age) during the first 18 months after FDA clearance. This analysis was undertaken with the goal of providing users and health care providers information they can use to obtain the best glycemic control from the iLet. Methods: Commercial users of the iLet who had a pre-iLet HbA1c value available and at least 3 weeks of iLet data in the cloud (n=5,987) were included in the analysis. Glucose management indicator (GMI) values and percentage of time in ranges (%70-180 mg/dl, %<70 mg/dl, and %<54 mg/dl) were calculated from all available CGM data. Users were classified according to the average number of meal announcements per day (MPD; from <0.5 to ≥4.5 per day in 0.5 meal increments) and separately according to the number of swipes to unlock the (STU; ≤4, 4-<13, and ≥13 swipes to unlock per day). Results: Mean baseline HbA1c was 8.6%; mean iLet GMI was 7.2%. Groups with <2.5 MPD had higher GMI and lower %70-180 mg/dl than those with ≥2.5 MPD. Groups with ≥2.5 to <4.5 MPD all had similar outcomes, with lower GMI and higher %70-180 mg/dl than groups with <2.5 MDP. There was little difference in %<70 mg/dl and %<54 mg/dl from <0.5 to <4.5 MPD. The ≥4.5 MPD group had lower GMI and higher %70-180 mg/dl than others, but it also had higher %<70 mg/dl and %<54 mg/dl. GMI and %<54 mg/dl was similar for all levels of STU. Higher STU was associated with higher %70-180 mg/dl and higher %<70 mg/dl. Discussion/Conclusion: In users of the iLet bionic pancreas the best outcomes were achieved by those announcing 2.5 to <4.5 MPD. This may correspond to typical numbers of meals and snacks consumed by users. Announcement of ≥4.5 MPD was associated with lower GMI and higher %70-180 mg/dl, but also higher %<54 mg/dl. Users announcing ≥4.5 MPD may be announcing “ghost” meals to obtain correction insulin, which risks “stacking” on top of iLet-determined correction doses and may lead to increased hypoglycemia. Higher STU was associated with higher %70-180 mg/dl and %<70 mg/dl but little difference in GMI and %<54 mg/dl.Presentation: Monday, July 14, 2025

Disclosure: A. Hanan: None. J.L. Snitzer: None.Introduction: Automated insulin delivery (AID) systems, such as Tandem Control-IQ, Medtronic 780G SmartGuard, and Omnipod Automated mode, have revolutionized diabetes management by providing real-time insulin adjustments based on continuous glucose monitoring (CGM) data. This report presents two distinct clinical cases that highlight the critical role of AID systems in managing diabetes mellitus (DM). Clinical Cases: The first case involves a 74-year-old male with Type 1 DM diagnosed in 1978, with complications including coronary artery disease, peripheral neuropathy, erectile dysfunction, and renal insufficiency. He maintained good glycemic control using a Tandem insulin pump and Dexcom CGM in Control-IQ mode despite not bolusing for food. However, after receiving a replacement pump, he failed to activate the Control-IQ feature, leading to significant deterioration of his glycemic metrics over two months. In September, with Control-IQ active 97% of the time, his average glucose was 187 mg/dL, time in range (TIR) 54%, standard deviation 88 mg/dL, glucose management indicator (GMI) 7.8%, and average daily insulin dose 37.56 units. By December, with Control-IQ inactive, his average glucose rose to 303 mg/dL, TIR dropped to 16%, standard deviation increased to 98 mg/dL, GMI rose to 10.6%, and his average daily insulin dose decreased to 25.16 units. The second case describes a 45-year-old male with Type 1 DM diagnosed in 1994, who had good diabetes control using a Medtronic 780G pump and Guardian 4 CGM. He inadvertently set his basal rate to zero but had the SmartGuard feature activated. Despite initial hyperglycemia with glucose levels in the 200-300 mg/dL range, the system was able to compensate, and after two weeks, it restored his glucose levels to his prior good control.Conclusion: These cases underscore the importance of AID systems in modern diabetes care. They demonstrate the importance of automation; the absence of Control-IQ in Case 1 revealed possible deficiencies in manual settings and lack of bolusing for food, leading to poor glycemic control. In Case 2, the Medtronic 780G SmartGuard worked well but if the patient went into manual mode, his BG’s would likely have been very high due to no basal rates. AID systems act as vital safety nets by reducing reliance on manual insulin adjustments and mitigating the impact of incorrect settings. Lastly, they highlight the critical need for patient education. Ensuring that patients understand how to turn on their AID function is essential, particularly when a pump is replaced. Overall, these findings emphasize the critical role of AID systems in enhancing glycemic control, mitigating human error, and improving the safety of diabetes management. They also highlight the need for comprehensive patient education, and periodic review with reassessment of pump settings to ensure optimal performance of these advanced technologies.Presentation: Monday, July 14, 2025

Disclosure: P.K. Patel: None. A. Saif: None. P.R. Schroeder: None. P.A. Sack: None.Introduction: Concentrated regular U-500 insulin (U-500R) is typically prescribed for patients with Type 2 diabetes mellitus (T2DM) who are very insulin-resistant requiring more than 200 units of insulin per day. U-500R can reduce the number of daily injections, lower the volume of injections, and enhance cost-effectiveness. Clinical trials have demonstrated improvements in glycemic control without an increased risk of hypoglycemia with the use of U-500R for continuous subcutaneous insulin infusion (CSII). However, automated insulin delivery (AID) systems, including hybrid closed-loop systems, have not yet been formally studied in T2DM patients using U-500R.Case: We present the case of a 39-year-old female with very insulin resistant T2DM who was initially treated with insulin lispro U100, using 100 units/day via the pump, which resulted in frequent pod changes every 1 to 1.5 days. In addition to the insulin she received from her pump, she also required ∼140 units of subcutaneous Basaglar daily. Her U100 lispro insulin, used via the pump, was switched to U-500R while continuing to use the automated mode and her Basaglar was discontinued. We maintained a 24-hour blood glucose target of 110 mg/dL and increased her active insulin time from 3 hours to 6 hours to better reflect the kinetics of U-500R. After four weeks, her average glucose level was 164 mg/dL. The distribution of her blood glucose (BG) readings was 1% <54 mg/dL, 1% < 70 mg/dL, 61% 70-180 mg/dL, 29% >180 mg/dL, and 8% > 250 mg/dL. The pump remained in auto mode 100% of the time. She was getting ∼38 units x 5 per day, or a total of 190 units. Of her total insulin usage, 59% was basal insulin and 41% was bolus insulin. Her insulin pod now lasts for 3 days. The HgA1C improved to 7.3 from 8.8 after two-month use of U500R in the Omnipod 5 AID system. She did not report any overt symptoms of hypoglycemia throughout this period, despite setting the BG target at the lowest possible level of 110. The rare episodes of hypoglycemia seemed to be mostly overnight. Conclusion: U-500R proved to be safe and effective with Omnipod 5 AID. Further research is required to adapt the automated dosing algorithm to U-500R to reduce the risk of nocturnal hypoglycemia.Presentation: Monday, July 14, 2025

Abstract Disclosure: S. Dinh: None. N. Sandhu: None. S. Mishra: None. Type 1 diabetes (T1D) is a chronic autoimmune disease that results in the destruction of insulin-secreting pancreatic beta-cells. Its early onset results in longer exposure to hyperglycemia and an increased risk of cardiovascular disease, the leading cause of death in patients with T1D. While the microvascular and macrovascular consequences of diabetes and hyperglycemia are well established and studied, majority of research focuses on the association between cardiovascular disease in patients with type 2 diabetes. Our study explores the clinical outcomes of patients with the less prevalent T1D and atrial fibrillation (AF). Patients with the primary diagnosis of AF were identified using the 2021 National Inpatient Sample and stratified based on the presence of T1D. The relationship between outcomes in patients with a history of T1D who were hospitalized for a primary diagnosis of AF was analyzed using multivariate regression analysis, adjusting for potential confounding variables. The threshold for statistical significance was set at a 2-tailed p-value of 0.05. In 2021, 369,399 adults were admitted with the primary diagnosis of AF. Of these, 49,444 had a concurrent diagnosis of T1D (13.4%). The mean age in the diabetes group was 70.8 years. Most patients in this group were male (52.3%), White (78.4%), had Medicare insurance (72.1%), and were admitted to urban teaching hospitals (49.1%). After adjusting for factors such as age, gender, insurance, income level, Charleston comorbidity index, location, and teaching status, there was decreased in-hospital mortality for patients admitted with AF who had concurrent T1D (aOR 0.64, 95% CI 0.48-0.84). This group also had decreased odds of vasopressor use (aOR 0.72, 95% CI 0.52-0.99), mechanical ventilation (aOR 0.72, 95% CI 0.55-0.96), and renal replacement therapy (aOR 0.09, 95% CI 0.05-0.15). Additionally, the African American population was found to have higher odds of mortality (aOR 1.34, 95% CI 1.03 - 1.74) and mechanical ventilation (aOR 1.79, 95% CI 1.41 - 2.29) compared to other groups. T1DM and AF are co-existing conditions that have been shown to have high mortality. However, our study shows that patients with T1DM who are hospitalized for AF may have better outcomes compared to those without T1DM. We speculate that the unexpected outcome difference could be a reflection of technology advancements like continuous glucose monitoring and automated insulin delivery systems for tight glycemic control. It also highlights racial disparities that providers should be aware of. Our study’s limitations, however, include potential misclassification bias with under or over-coding, inability to account for longer-term outcomes such as readmission, and other unaccounted confounding variables. Presentation: Saturday, July 12, 2025

Physical activity plays a critical role in reducing cardiovascular risk among individuals with type 1 diabetes (PwT1D); however, more than two-thirds of PwT1D remain insufficiently active, primarily due to the unpredictability of glucose fluctuations and the fear of hypoglycaemia during exercise. Automated insulin delivery systems, such as the MiniMed™ 780G (MM780G), offer potential solutions by automating insulin dosing, supporting stable glucose control, and allowing for a higher temporary glucose target. This narrative review summarizes the current evidence on the safety and efficacy of the MM780G system during exercise in PwT1D, with a focus on the temporary target (TT) feature and real-world usability. The MM780G consistently supports glycaemic safety during exercise, especially moderate-intensity aerobic activity, with time in range frequently meeting or exceeding consensus guidelines. Activation of the TT feature 60- to 120-min pre-exercise, particularly when paired with reduced meal boluses, significantly lowers the risk of hypoglycaemia. Real-world studies confirm its safety across varied activities, including prolonged aerobic exercise and contact sports. While evidence for resistance and high-intensity exercise remains limited, initial findings are favourable. User education is crucial to optimize use, particularly in competitive and prolonged settings. The MM780G is a safe and effective tool for glycaemic management during exercise in PwT1D. Best outcomes are achieved through strategic use of the TT feature and meal bolus reduction, individualized carbohydrate strategies and proper education. Applying easy-to-follow pre-exercising recommendations may encourage better engagement with regular exercising. Further research is needed on long-term adaptations to training and resistance-based exercise protocols.

Physical activity (PA) represents a significant challenge in the management of type 1 diabetes (T1D), given its impact on glucose levels, which are influenced by various exercise characteristics, including duration, intensity, and type. The development of strategies that allow for the monitoring of these characteristics is crucial to improve glycemic control during exercise for both conventional therapies and automated insulin delivery systems. This paper presents a state-space model that further exploits the HR signal for the purpose of quantifying and distinguishing between aerobic and anaerobic PA. The model design is based on an analysis of the distinctive features of HR signal, including the mean HR value, the maximum HR, and the presence of pronounced fluctuations in HR. This method does not require any training and offers users interpretability and explainability. Furthermore, it enables intuitive tuning, a feature which is of particular importance in clinical settings. The model is validated using two clinical trials: the T1DEXI study, which is the largest real-world clinical trial including PA in people with T1D conducted to date, and a pilot clinical trial conducted by our research group in Argentina. The findings indicate the model has the capacity to quantify and differentiate between aerobic and resistance PA, which represent the two types of PA exhibiting the most significant and contrasting influence on glucose levels.