Advanced Closed-Loop Control System Improves Postprandial Glycemic Control Compared With a Hybrid Closed-Loop System Following Unannounced Meal

Use of diabetes technology in children

Children with type 1 diabetes and their caregivers face numerous challenges navigating the unpredictability of this complex disease. Although the burden of managing diabetes remains significant, new technology has eased some of the load and allowed children with type 1 diabetes to achieve tighter glycaemic management without fear of excess hypoglycaemia. Continuous glucose monitor use alone improves outcomes and is considered standard of care for paediatric type 1 diabetes management. Similarly, automated insulin delivery (AID) systems have proven to be safe and effective for children as young as 2 years of age. AID use improves not only blood glucose levels but also quality of life for children with type 1 diabetes and their caregivers and should be strongly considered for all youth with type 1 diabetes if available and affordable. Here, we review key data on the use of diabetes technology in the paediatric population and discuss management issues unique to children and adolescents.

Time in Tight Glucose Range in Type 1 Diabetes: Predictive Factors and Achievable Targets in Real-World Users of the MiniMed 780G System

OBJECTIVE

We studied time in tight range (TITR) (70-140 mg/dL) in real-world users of the MiniMed 780G system (MM780G).

RESEARCH DESIGN AND METHODS

CareLink Personal data were extracted (August 2020 to December 2022) to examine TITR and its relationship with time in range (TIR; 70-180 mg/dL), factors predicting higher TITR, and which TITR target is a reasonable treatment goal.

RESULTS

The 13,461 users (3,762 age ≤15 years and 9,699 age >15 years) showed an average TITR of 48.9% in those age ≤15 years and 48.8% in the older group (vs. TIR 71.2% and 73.9%, respectively). Consistent use of a glucose target (GT) of 100 mg/dL and active insulin time (AIT) of 2 h were the most relevant factors predicting higher TITR (P < 0.0001). In users consistently applying these optimal settings, TITR was 56.7% in those age ≤15 years and 57.0% in the older group, and the relative impact of these settings on TITR was 60% and 86% greater than that on TIR, respectively. TITRs of ∼45% (age ≤15 years 46.3% and older group 45.4%), ∼50% (50.7% and 50.7%) and ∼55% (56.4% and 58.0%) were best associated with glucose management indicators <7.0%, <6.8%, and <6.5%, respectively. TITRs of >45%, >50%, and >55% were achieved in 91%, 74%, and 55% of those age ≤15 years and 93%, 81%, and 57% of older group users, respectively, at optimal settings.

CONCLUSIONS

This study demonstrates that 1) mean TIR is high with a high mean TITR in MM780G users (>48%), 2) consistent use of optimal GT/AIT improves TITR (>56%), 3) the impact of these settings on TITR is larger than on TIR, and 4) a TITR target >50% is our suggested treatment goal.

An ensemble machine learning approach for the detection of unannounced meals to enhance postprandial glucose control

BACKGROUND

Hybrid automated insulin delivery systems enhance postprandial glucose control in type 1 diabetes, however, meal announcements are burdensome. To overcome this, we propose a machine learning-based automated meal detection approach; METHODS:: A heterogeneous ensemble method combining an artificial neural network, random forest, and logistic regression was employed. Trained and tested on data from two in-silico cohorts comprising 20 and 47 patients. It accounted for various meal sizes (moderate to high) and glucose appearance rates (slow and rapid absorbing). To produce an optimal prediction model, three ensemble configurations were used: logical AND, majority voting, and logical OR. In addition to the in-silico data, the proposed meal detector was also trained and tested using the OhioT1DM dataset. Finally, the meal detector is combined with a bolus insulin compensation scheme; RESULTS:: The ensemble majority voting obtained the best meal detector results for both the in-silico and OhioT1DM cohorts with a sensitivity of 77%, 94%, 61%, precision of 96%, 89%, 72%, F1-score of 85%, 91%, 66%, and with false positives per day values of 0.05, 0.19, 0.17, respectively. Automatic meal detection with insulin compensation has been performed in open-loop insulin therapy using the AND ensemble, chosen for its lower false positive rate. Time-in-range has significantly increased 10.48% and 16.03%, time above range was reduced by 5.16% and 11.85%, with a minimal time below range increase of 0.35% and 2.69% for both in-silico cohorts, respectively, compared to the results without a meal detector; CONCLUSION:: To increase the overall accuracy and robustness of the predictions, this ensemble methodology aims to take advantage of each base model's strengths. All of the results point to the potential application of the proposed meal detector as a separate module for the detection of meals in automated insulin delivery systems to achieve improved glycemic control.

Evidence from clinical trials on high-risk medical devices in children: a scoping review

BACKGROUND

Meeting increased regulatory requirements for clinical evaluation of medical devices marketed in Europe in accordance with the Medical Device Regulation (EU 2017/745) is challenging, particularly for high-risk devices used in children.

METHODS

Within the CORE-MD project, we performed a scoping review on evidence from clinical trials investigating high-risk paediatric medical devices used in paediatric cardiology, diabetology, orthopaedics and surgery, in patients aged 0-21 years. We searched Medline and Embase from 1st January 2017 to 9th November 2022.

RESULTS

From 1692 records screened, 99 trials were included. Most were multicentre studies performed in North America and Europe that mainly had evaluated medical devices from the specialty of diabetology. Most had enrolled adolescents and 39% of trials included both children and adults. Randomized controlled trials accounted for 38% of the sample. Other frequently used designs were before-after studies (21%) and crossover trials (20%). Included trials were mainly small, with a sample size <100 participants in 64% of the studies. Most frequently assessed outcomes were efficacy and effectiveness as well as safety.

CONCLUSION

Within the assessed sample, clinical trials on high-risk medical devices in children were of various designs, often lacked a concurrent control group, and recruited few infants and young children.

IMPACT

In the assessed sample, clinical trials on high-risk medical devices in children were mainly small, with variable study designs (often without concurrent control), and they mostly enrolled adolescents. We provide a systematic summary of methodologies applied in clinical trials of medical devices in the paediatric population, reflecting obstacles in this research area that make it challenging to conduct adequately powered randomized controlled trials. In view of changing European regulations and related concerns about shortages of high-risk medical devices for children, our findings may assist competent authorities in setting realistic requirements for the evidence level to support device conformity certification.

Artificial Intelligence and Machine Learning for Improving Glycemic Control in Diabetes: Best Practices, Pitfalls, and Opportunities

OBJECTIVE

Artificial intelligence and machine learning are transforming many fields including medicine. In diabetes, robust biosensing technologies and automated insulin delivery therapies have created a substantial opportunity to improve health. While the number of manuscripts addressing the topic of applying machine learning to diabetes has grown in recent years, there has been a lack of consistency in the methods, metrics, and data used to train and evaluate these algorithms. This manuscript provides consensus guidelines for machine learning practitioners in the field of diabetes, including best practice recommended approaches and warnings about pitfalls to avoid.

METHODS

Algorithmic approaches are reviewed and benefits of different algorithms are discussed including importance of clinical accuracy, explainability, interpretability, and personalization. We review the most common features used in machine learning applications in diabetes glucose control and provide an open-source library of functions for calculating features, as well as a framework for specifying data sets using data sheets. A review of current data sets available for training algorithms is provided as well as an online repository of data sources.

SIGNIFICANCE

These consensus guidelines are designed to improve performance and translatability of new machine learning algorithms developed in the field of diabetes for engineers and data scientists.

NIH Fifth Artificial Pancreas Workshop 2023: Meeting Report: The Fifth Artificial Pancreas Workshop: Enabling Fully Automation, Access, and Adoption

The Fifth Artificial Pancreas Workshop: Enabling Fully Automation, Access, and Adoption was held at the National Institutes of Health (NIH) Campus in Bethesda, Maryland on May 1 to 2, 2023. The organizing Committee included representatives of NIH, the US Food and Drug Administration (FDA), Diabetes Technology Society, Juvenile Diabetes Research Foundation (JDRF), and the Leona M. and Harry B. Helmsley Charitable Trust. In previous years, the NIH Division of Diabetes, Endocrinology, and Metabolic Diseases along with other diabetes organizations had organized periodic workshops, and it had been seven years since the NIH hosted the Fourth Artificial Pancreas in July 2016. Since then, significant improvements in insulin delivery have occurred. Several automated insulin delivery (AID) systems are now commercially available. The workshop featured sessions on: (1) Lessons Learned from Recent Advanced Clinical Trials and Real-World Data Analysis, (2) Interoperability, Data Management, Integration of Systems, and Cybersecurity, Challenges and Regulatory Considerations, (3) Adaptation of Systems Through the Lifespan and Special Populations: Are Specific Algorithms Needed, (4) Development of Adaptive Algorithms for Insulin Only and for Multihormonal Systems or Combination with Adjuvant Therapies and Drugs: Clinical Expected Outcomes and Public Health Impact, (5) Novel Artificial Intelligence Strategies to Develop Smarter, More Automated, Personalized Diabetes Management Systems, (6) Novel Sensing Strategies, Hormone Formulations and Delivery to Optimize Close-loop Systems, (7) Special Topic: Clinical and Real-world Viability of IP-IP Systems. "Fully automated closed-loop insulin delivery using the IP route," (8) Round-table Panel: Closed-loop performance: What to Expect and What are the Best Metrics to Assess it, and (9) Round-table Discussion: What is Needed for More Adaptable, Accessible, and Usable Future Generation of Systems? How to Promote Equitable Innovation? This article summarizes the discussions of the Workshop.

Detection of Meals and Physical Activity Events From Free-Living Data of People With Diabetes

BACKGROUND

Predicting carbohydrate intake and physical activity in people with diabetes is crucial for improving blood glucose concentration regulation. Patterns of individual behavior can be detected from historical free-living data to predict meal and exercise times. Data collected in free-living may have missing values and forgotten manual entries. While machine learning (ML) can capture meal and exercise times, missing values, noise, and errors in data can reduce the accuracy of ML algorithms.

METHODS

Two recurrent neural networks (RNNs) are developed with original and imputed data sets to assess detection accuracy of meal and exercise events. Continuous glucose monitoring (CGM) data, insulin infused from pump data, and manual meal and exercise entries from free-living data are used to predict meals, exercise, and their concurrent occurrence. They contain missing values of various lengths in time, noise, and outliers.

RESULTS

The accuracy of RNN models range from 89.9% to 95.7% for identifying the state of event (meal, exercise, both, or neither) for various users. "No meal or exercise" state is determined with 94.58% accuracy by using the best RNN (long short-term memory [LSTM] with 1D Convolution). Detection accuracy with this RNN is 98.05% for meals, 93.42% for exercise, and 55.56% for concurrent meal-exercise events.

CONCLUSIONS

The meal and exercise times detected by the RNN models can be used to warn people for entering meal and exercise information to hybrid closed-loop automated insulin delivery systems. Reliable accuracy for event detection necessitates powerful ML and large data sets. The use of additional sensors and algorithms for detecting these events and their characteristics provides a more accurate alternative.

Performance Effect of Adjusting Insulin Sensitivity for Model-Based Automated Insulin Delivery Systems

BACKGROUND

Model predictive control (MPC) has become one of the most popular control strategies for automated insulin delivery (AID) in type 1 diabetes (T1D). These algorithms rely on a prediction model to determine the best insulin dosing every sampling time. Although these algorithms have been shown to be safe and effective for glucose management through clinical trials, managing the ever-fluctuating relationship between insulin delivery and resulting glucose uptake (aka insulin sensitivity, IS) remains a challenge. We aim to evaluate the effect of informing an AID system with IS on the performance of the system.

METHOD

The University of Virginia (UVA) MPC control-based hybrid closed-loop (HCL) and fully closed-loop (FCL) system was used. One-day simulations at varying levels of IS were run with the UVA/Padova T1D Simulator. The AID system was informed with an estimated value of IS obtained through a mixed meal glucose tolerance test. Relevant controller parameters are updated to inform insulin dosing of IS. Performance of the HCL/FCL system with and without information of the changing IS was assessed using a novel performance metric penalizing the time outside the target glucose range.

RESULTS

Feedback in AID systems provides a certain degree tolerance to changes in IS. However, IS-informed bolus and basal dosing improve glycemic outcomes, providing increased protection against hyperglycemia and hypoglycemia according to the individual's physiological state.

CONCLUSIONS

The proof-of-concept analysis presented here shows the potentially beneficial effects on system performance of informing the AID system with accurate estimates of IS. In particular, when considering reduced IS, the informed controller provides increased protection against hyperglycemia compared with the naïve controller. Similarly, reduced hypoglycemia is obtained for situations with increased IS. Further tailoring of the adaptation schemes proposed in this work is needed to overcome the increased hypoglycemia observed in the more resistant cases and to optimize the performance of the adaptation method.

Multivariable Automated Insulin Delivery System for Handling Planned and Spontaneous Physical Activities

BACKGROUND

Hybrid closed-loop control of glucose levels in people with type 1 diabetes mellitus (T1D) is limited by the requirements on users to manually announce physical activity (PA) and meals to the artificial pancreas system. Multivariable automated insulin delivery (mvAID) systems that can handle unannounced PAs and meals without any manual announcements by the user can improve glycemic control by modulating insulin dosing in response to the occurrence and intensity of spontaneous physical activities.

METHODS

An mvAID system is developed to supplement the glucose measurements with additional physiological signals from a wristband device, with the signals analyzed using artificial intelligence algorithms to automatically detect the occurrence of PA and estimate its intensity. This additional information gained from the physiological signals enables more proactive insulin dosing adjustments in response to both planned exercise and spontaneous unanticipated physical activities.

RESULTS

In silico studies of the mvAID illustrate the safety and efficacy of the system. The mvAID is translated to pilot clinical studies to assess its performance, and the clinical experiments demonstrate an increased time in range and reduced risk of hypoglycemia following unannounced PA and meals.

CONCLUSIONS

The mvAID systems can increase the safety and efficacy of insulin delivery in the presence of unannounced physical activities and meals, leading to improved lives and less burden on people with T1D.

Assessment of Meal Anticipation for Improving Fully Automated Insulin Delivery in Adults With Type 1 Diabetes

OBJECTIVE

Meals are a consistent challenge to glycemic control in type 1 diabetes (T1D). Our objective was to assess the glycemic impact of meal anticipation within a fully automated insulin delivery (AID) system among adults with T1D.

RESEARCH DESIGN AND METHODS

We report the results of a randomized crossover clinical trial comparing three modalities of AID systems: hybrid closed loop (HCL), full closed loop (FCL), and full closed loop with meal anticipation (FCL+). Modalities were tested during three supervised 24-h admissions, where breakfast, lunch, and dinner were consumed per participant's home schedule, at a fixed time, and with a 1.5-h delay, respectively. Primary outcome was the percent time in range 70-180 mg/dL (TIR) during the breakfast postprandial period for FCL+ versus FCL.

RESULTS

Thirty-five adults with T1D (age 44.5 ± 15.4 years; HbA1c 6.7 ± 0.9%; n = 23 women and n = 12 men) were randomly assigned. TIR for the 5-h period after breakfast was 75 ± 23%, 58 ± 21%, and 63 ± 19% for HCL, FCL, and FCL+, respectively, with no significant difference between FCL+ and FCL. For the 2 h before dinner, time below range (TBR) was similar for FCL and FCL+. For the 5-h period after dinner, TIR was similar for FCL+ and FCL (71 ± 34% vs. 72 ± 29%; P = 1.0), whereas TBR was reduced in FCL+ (median 0% [0-0%] vs. 0% [0-0.8%]; P = 0.03). Overall, 24-h control for HCL, FCL, and FCL+ was 86 ± 10%, 77 ± 11%, and 77 ± 12%, respectively.

CONCLUSIONS

Although postprandial control remained optimal with hybrid AID, both fully AID solutions offered overall TIR >70% with similar or lower exposure to hypoglycemia. Anticipation did not significantly improve postprandial control in AID systems but also did not increase hypoglycemic risk when meals were delayed.

Efficacy and safety of closed-loop control system for type one diabetes in adolescents a meta analysis

This meta-analysis compares the efficacy and safety of Closed-Loop Control (CLC) to Sensor-Augmented Insulin Pump (SAP) for adolescent patients with Type 1 Diabetes Mellitus (T1DM). Eleven randomized-controlled trials were included with a total of 570 patients, from a total of 869 articles found adhering to PRISMA guidelines. The efficacy of the therapies were evaluated from the day, night and during physical activities monitoring of the of the mean blood glucose (BG), Time In Range (TIR), and Standard Deviation (SD) of the glucose variability. The safety measure of the therapies, was assessed from the day and night recording of the hypoglycemic and hyperglycemic events occurred. Pooled results of comparison of mean BG values for day, night and physical activities, - 4.33 [- 6.70, - 1.96] (P = 0.0003), - 16.61 [- 31.68, - 1.54] (P = 0.03) and - 8.27 [- 19.52, 2.99] (P = 0.15). The monitoring for day, night and physical activities for TIR - 13.18 [- 19.18, - 7.17] (P < 0.0001), - 15.36 [- 26.81, - 3.92] (P = 0.009) and - 7.39 [- 17.65, 2.87] (P = 0.16). The day and night results of SD of glucose variability was - 0.40 [- 0.79, - 0.00] (P = 0.05) and - 0.86 [- 2.67, 0.95] (P = 0.35). These values shows the superiority of CLC system in terms of efficacy. The safety evaluation, of the day, night and physical activities observations of average blood glucose goal hypoglycemic events - 0.54 [- 1.86, 0.79] (P = 0.43), 0.04 [- 0.20, 0.27] (P = 0.77) and 0.00 [- 0.25, 0.25] (P = 1.00) and hyperglycemic events - 0.04 [- 0.20, 0.27] (P = 0.77), - 7.11 [- 12.77, - 1.45] (P = 0.01) and - 0.00 [- 0.10, 0.10] (P = 0.97), highlights the commendable safety factor of CLC. The CLC systems can be considered as an ideal preference in the management of adolescents with type 1 diabetes to be used during a 24 h basis.

Updates on Glycaemic Control Strategies: A Range of Opportunities after Total Pancreatectomy

In the past, indications for total pancreatectomy (TP) were rare, with several concerns about patients' postoperative quality of life due to exocrine and endocrine post-pancreatectomy management [...].

Diabetes Technology Meeting 2021

Diabetes Technology Society hosted its annual Diabetes Technology Meeting on November 4 to November 6, 2021. This meeting brought together speakers to discuss various developments within the field of diabetes technology. Meeting topics included blood glucose monitoring, continuous glucose monitoring, novel sensors, direct-to-consumer telehealth, metrics for glycemia, software for diabetes, regulation of diabetes technology, diabetes data science, artificial pancreas, novel insulins, insulin delivery, skin trauma, metabesity, precision diabetes, diversity in diabetes technology, use of diabetes technology in pregnancy, and green diabetes. A live demonstration on a mobile app to monitor diabetic foot wounds was presented.

Automated Insulin Delivery with SGLT2i Combination Therapy in Type 1 Diabetes

Background: Use of sodium-glucose cotransporter 2 inhibitors (SGLT2i) as adjunct therapy to insulin in type 1 diabetes (T1D) has been previously studied. In this study, we present data from the first free-living trial combining low-dose SGLT2i with commercial automated insulin delivery (AID) or predictive low glucose suspend (PLGS) systems. Methods: In an 8-week, randomized, controlled crossover trial, adults with T1D received 5 mg/day empagliflozin (EMPA) or no drug (NOEMPA) as adjunct to insulin therapy. Participants were also randomized to sequential orders of AID (Control-IQ) and PLGS (Basal-IQ) systems for 4 and 2 weeks, respectively. The primary endpoint was percent time-in-range (TIR) 70-180 mg/dL during daytime (7:00-23:00 h) while on AID (NCT04201496). Findings: A total of 39 subjects were enrolled, 35 were randomized, 34 (EMPA; n = 18 and NOEMPA n = 16) were analyzed according to the intention-to-treat principle, and 32 (EMPA; n = 16 and NOEMPA n = 16) completed the trial. On AID, EMPA versus NOEMPA had higher daytime TIR 81% versus 71% with a mean estimated difference of +9.9% (confidence interval [95% CI] 0.6-19.1); p = 0.04. On PLGS, the EMPA versus NOEMPA daytime TIR was 80% versus 63%, mean estimated difference of +16.5% (95% CI 7.3-25.7); p < 0.001. One subject on SGLT2i and AID had one episode of diabetic ketoacidosis with nonfunctioning insulin pump infusion site occlusion contributory. Interpretation: In an 8-week outpatient study, addition of 5 mg daily empagliflozin to commercially available AID or PLGS systems significantly improved daytime glucose control in individuals with T1D, without increased hypoglycemia risk. However, the risk of ketosis and ketoacidosis remains. Therefore, future studies with SGLT2i will need modifications to closed-loop control algorithms to enhance safety.

Carbohydrate Tolerance Threshold for Unannounced Snacks in Children and Adolescents With Type 1 Diabetes Using an Advanced Hybrid Closed-Loop System

OBJECTIVE

To find a carbohydrate (CHO) tolerance threshold for unannounced snacks to avoid the 2 h increase in glycemia (difference between pre- and postmeal blood glucose [ΔBG]) ≥50 mg/dL in advanced hybrid closed-loop (a-HCL) users.

RESEARCH DESIGN AND METHODS

Fourteen children and adolescents with type 1 diabetes (7 females; mean age [± SD] 14.5 ± 3.6 years), users of the Medtronic MiniMed 780G, participated in the study. For 12 days, they did not perform insulin bolus before breakfasts, with defined different quantities and types of CHO, with or without fats, performing blood glucose (BG) before and 2 h after the meal.

RESULTS

A cutoff of 19.8 g of total CHO was found to determine a ΔBG of 50 mg/dL. BG never exceeded 250 mg/dL. Mean time in range was ≥70% in the 2 h following each snack.

CONCLUSIONS

Unannounced snacks of up to 20 g of CHO can avoid ΔBG ≥50 mg/dL in MiniMed 780G users, although unannounced meals of up to 30 g of CHO are safe.

A Deep Learning Framework for Automatic Meal Detection and Estimation in Artificial Pancreas Systems

Current artificial pancreas (AP) systems are hybrid closed-loop systems that require manual meal announcements to manage postprandial glucose control effectively. This poses a cognitive burden and challenge to users with T1D since this relies on frequent user engagement to maintain tight glucose control. In order to move towards fully automated closed-loop glucose control, we propose an algorithm based on a deep learning framework that performs multitask quantile regression, for both meal detection and carbohydrate estimation. Our proposed method is evaluated in silico on 10 adult subjects from the UVa/Padova simulator with a Bio-inspired Artificial Pancreas (BiAP) control algorithm over a 2 month period. Three different configurations of the AP are evaluated -BiAP without meal announcement (BiAP-NMA), BiAP with meal announcement (BiAP-MA), and BiAP with meal detection (BiAP-MD). We present results showing an improvement of BiAP-MD over BiAP-NMA, demonstrating 144.5 ± 6.8 mg/dL mean blood glucose level (-4.4 mg/dL, p< 0.01) and 77.8 ± 6.3% mean time between 70 and 180 mg/dL (+3.9%, p< 0.001). This improvement in control is realised without a significant increase in mean in hypoglycaemia (+0.1%, p= 0.4). In terms of detection of meals and snacks, the proposed method on average achieves 93% precision and 76% recall with a detection delay time of 38 ± 15 min (92% precision, 92% recall, and 37 min detection time for meals only). Furthermore, BiAP-MD handles hypoglycaemia better than BiAP-MA based on CVGA assessment with fewer control errors (10% vs. 20%). This study suggests that multitask quantile regression can improve the capability of AP systems for postprandial glucose control without increasing hypoglycaemia.