On the road to fully automated insulin delivery: A systematic review of meal announcement free algorithms

From algorithm to alarm: a narrative review on insulin delivery innovation

Background and objective: The global rise in the incidence of diabetes mellitus necessitates innovative management strategies to mitigate the severity of the disease. The mainstay of diabetes treatment is still insulin therapy, although there are drawbacks, such as the need to maintain ideal glycemic control while lowering the risk of hypoglycemia and user fatigue. The development, clinical usefulness, and constraints of insulin delivery systems are examined in this narrative review, with particular attention to sensor-augmented and closed-loop technologies. Methods: A thorough electronic database search was conducted to obtain pertinent data on major insulin delivery devices, including sensor-augmented and closed-loop systems. We studied advancements in insulin delivery systems over time and their impact on the management of diabetes mellitus. We thoroughly examined both qualitative and quantitative data to summarize our conclusions. Key content and findings: Closed-loop insulin delivery systems, which combine insulin pumps, advanced control algorithms, and continuous glucose monitoring, are the next step forward in the treatment of diabetes. These systems autonomously regulate blood glucose levels, reducing the need for user intervention. Closed-loop devices outperform sensor-augmented pump (SAP) treatment in reducing hypoglycemia, extending the time-in-range, and lowering self-management responsibilities. In order to provide real-time control, closed-loop systems dynamically modify insulin delivery in response to glucose variations by using algorithms powered by artificial intelligence. Because SAP therapy depends more on user input and may not be as successful in providing consistent glycemic control, this capability helps overcome its drawbacks. Further improving overall safety and efficacy, the closed-loop technique decreases blood glucose variability. Conclusion: The two primary developments in diabetes care are sensor-augmented and closed-loop insulin devices, which offer better glucose control and lower hypoglycemia rates compared to traditional therapies. Closed-loop insulin delivery systems seem to be a more clinically successful option for controlling diabetes mellitus than SAP therapy.

[Continuous glucose monitoring and automated insulin dosing systems in clinical routine].

Multicenter Comparison of Computerized Insulin Sensitivity Coefficient and Paper-Based Intravenous Insulin Protocols.

INTRODUCTION Computerized intravenous (IV) insulin infusion calculators may offer a safe and efficacious means of achieving glycemic control. Historically, our health system utilized various paper-based IV insulin infusion protocols but transitioned to a computerized insulin infusion calculator protocol based on the insulin sensitivity coefficient (ISC) in 2024. The objective of this study was to compare the safety and efficacy of the computerized ISC-based insulin infusion calculator protocol with previous protocols. METHODS This was a pre-post intervention analysis of critically ill adult patients initiated on a paper-based insulin infusion across 13 institutions from March 2023 to May 2023 for pre-analysis and patients on a computerized, ISC-based insulin infusion from November 2024 to December 2024 for post-analysis. The primary outcome was the percentage of blood glucose levels within target range (80-180 mg/dL). Secondary outcomes included rates of hyperglycemia (>200 mg/dL), severe hyperglycemia (>250 mg/dL), hypoglycemia (<70 mg/dL), and severe hypoglycemia (<54 mg/dL). RESULTS In total, 207 and 237 patients were included in the pre- and post-intervention groups, respectively. For the primary outcome, 4610 of 6313 (73%) and 6163 of 7870 (78.3%) of blood glucose values were in range in the pre- and post-intervention groups, respectively (P < .01). There was no difference in severe hypoglycemia (0.2% vs 0.1%, P = .24), but rates of hyperglycemia (16.6% vs 12.6%, P < .01), severe hyperglycemia (5.7% vs 4.0%, P < .01), and hypoglycemia (0.7% vs 0.2%, P < .01) were higher in the pre-intervention group. CONCLUSIONS A computerized ISC-based insulin infusion protocol resulted in more blood glucose values in target range and lower rates of hypoglycemia compared with paper-based protocols.

Simplifying meal bolusing with Omnipod 5: real-world outcomes to support simplified user-initiated insulin dosing strategies in managing type 1 or type 2 diabetes.

AIMS Automated insulin delivery (AID) systems typically require user‑initiated meal boluses. This study leveraged real-world data from Omnipod® 5 Automated Insulin Delivery System users to evaluate whether glycemic outcomes using simplified meal bolusing were non-inferior to carbohydrate counting. METHODS A retrospective analysis of Omnipod 5 users in the US, UK, France, and Germany from 01/01/25 - 31/03/25 was conducted. Carbohydrate counting or simplified bolusing categorization was based on the proportion of boluses that used each individual's top five carbohydrate entries. The primary comparison used quantile regression and adjusted for users' characteristics and device settings/usage. RESULTS The overall cohort included 195,529 individuals (92 % T1D). In T1D, simplified bolusing was associated with 2.5 % lower time in range (TIR) 70-180 mg/dL, 0.19 % lower time below range (TBR) < 70 mg/dL, and 5 mg/dL higher mean glucose compared with carbohydrate counting. All endpoints met noninferiority, with similar findings across age groups and countries. In T2D, differences were - 1.7 % for TIR, -0.05 % for TBR, and + 4 mg/dL for mean glucose, also meeting noninferiority. Among those using optimized settings and expected use patterns, T1D maintained noninferiority and T2D met noninferiority for TBR only. CONCLUSIONS Glycemic outcomes with simplified bolusing were non‑inferior to carbohydrate counting, suggesting carbohydrate counting ability should not affect AID access.

Early Consideration of Automated Insulin Delivery in Type 2 Diabetes: A Critical Review.

Despite major advances in therapeutics, many adults with type 2 diabetes do not achieve recommended glycemic targets, driven by progressive beta cell dysfunction and persistent therapeutic inertia. Insulin therapy is commonly delayed for years, exposing patients to prolonged hyperglycemia and substantially increasing complication risks. Although automated insulin delivery (AID) systems have transformed outcomes for people with type 1 diabetes, their adoption in type 2 diabetes has been limited. Emerging evidence and recent updates to the American Diabetes Association Standards of Care now support AID as the preferred insulin delivery method for adults with type 2 diabetes on multiple daily injections of insulin or for whom basal-only insulin is not sufficient to achieve glycemic targets, reflecting a paradigm shift toward addressing the burdens and limitations of manual insulin delivery. Randomized and real-world studies demonstrate that AID systems significantly improve A1C and time in range while minimizing hypoglycemia, without increasing treatment complexity or patient burden. By dynamically adapting insulin delivery to real-time continuous glucose monitoring data, AID mitigates common barriers to intensification, including fear of hypoglycemia and increased regimen complexity associated with multiple daily injections. In addition, AID enables more efficient diabetes care through standardized glucose reporting, remote data review, and team-based management. Together, these advances position AID as a practical and effective strategy to overcome long-standing therapeutic inertia in type 2 diabetes. Earlier consideration of AID during insulin intensification could improve glycemic outcomes, reduce diabetes-related complications, decrease cognitive burden, and modernize insulin delivery for people living with type 2 diabetes.

Reinforcement Learning for Automated Insulin Delivery: A Comprehensive Review of Methods, Challenges, and Future Directions

Type 1 Diabetes Mellitus (T1DM) is a chronic autoimmune disease characterized by the destruction of pancreatic β cells, resulting in complete insulin deficiency and an exogenous insulin requirement. However, despite the technological progress, classical insulin therapies and control algorithms such as Proportional-Integral-Derivative (PID) and Model Predictive Control (MPC) are not able to follow unpredictable glucose variations due to meals, exercise, and stress. Automated Insulin Delivery (AID) systems combine continuous glucose monitoring, insulin pumps, and algorithmic decision-making but remain limited by model dependency, computational load, and lack of personalization. Reinforcement Learning (RL) has emerged as a transformative paradigm for AI systems by modeling insulin regulation as a sequential decision-making process that learns optimal dosing policies from experience. Model-free RL algorithms like Deep Q-Networks (DQN) and Proximal Policy Optimization (PPO) have demonstrated improved glycemic control, whereas model-based and hybrid RL algorithms have demonstrated increased safety and sample efficiency by integrating physiological constraints. Deep Reinforcement Learning (DRL) further increases adaptability through neural network-based personalization over different patient profiles. RL can be integrated with wearable technologies, Internet of Things (IoT) ecosystems, and digital twin simulations for continuous learning, remote monitoring, and patient specific optimization. Early clinical studies show improved time-in-range (TIR), reduced hypoglycemia, and improved long-term glucose control compared to traditional methods. However, there are still several challenges, including lack of real-world data, safety in online learning, interpretability, and regulatory compliance. Future advances will need to include the development of explainable and federated RL frameworks, lightweight models for wearable deployment, and multi-hormone control systems. In summary, RL-based AID systems represent an important step towards fully autonomous and personalized artificial pancreas solutions. Their clinical success will be dependent on multidisciplinary collaboration between AI researchers, clinicians, and regulators to ensure safety, transparency, and reliability in next generation diabetes management.