Arguably, diabetes mellitus is one of the best-quantified human conditions: elaborate in silico models describe the action of the human metabolic system; real-time signals such as continuous glucose ...monitoring are readily available; insulin delivery is being automated; and control algorithms are capable of optimizing blood glucose fluctuation in patients’ natural environments. The transition of the artificial pancreas (AP) to everyday clinical use is happening now, and is contingent upon seamless concerted work of devices encompassing the patient in a digital treatment ecosystem. This review recounts briefly the story of diabetes technology, which began a century ago with the discovery of insulin, progressed through glucose monitoring and subcutaneous insulin delivery, and is now rapidly advancing towards fully automated clinically viable AP systems.
Classic studies have shown that the complications of diabetes can be reduced by strict glycemic control. However, the risk of hypoglycemia remains the primary barrier to intensive therapy; thus, people with diabetes face a life-long optimization problem: reduce their HbA1c while simultaneously avoiding hypoglycemia.Contemporary technologies, such as CGM and CLC (known as the AP), offer the best solution to this optimization problem, particularly in type 1 diabetes.The advancement of these technologies is enabled by decades of studies aiming to develop elaborate in silico models describing the action of the human metabolic system, process CGM data and other metabolic signals, and automate insulin delivery.As a result, 100 years after the discovery of insulin, the technology is entering the stage of fully automated portable AP systems providing real-time, long-term optimal control of diabetes in patients’ natural environments.
Background:
Adaptive model predictive control (MPC) algorithms that recursively update the glucose prediction model are shown to be promising in the development of fully automated multivariable ...artificial pancreas systems. However, the recursively updated glycemic prediction models do not explicitly consider prior knowledge in the identification of the model parameters. Prior information of the glycemic effects of meals and physical activity can improve model accuracy and yield better glycemic control algorithms.
Methods:
A glucose prediction model based on regularized partial least squares (rPLS) method where the prior information is encoded as the regularization term is developed to provide accurate predictions of the future glucose concentrations. An adaptive MPC is developed that incorporates dynamic trajectories for the glucose setpoint and insulin dosing constraints based on the estimated plasma insulin concentration (PIC). The proposed adaptive MPC algorithm is robust to disturbances caused by unannounced meals and physical activities even in cases with missing glucose measurements. The effectiveness of the proposed adaptive MPC based on rPLS is investigated with in silico subjects of the multivariable glucose-insulin-physiological variables simulator (mGIPsim).
Results:
The efficacy of the proposed adaptive MPC strategy in regulating the blood glucose concentration (BGC) of people with T1DM is assessed using the average percent time in range (TIR) for glucose, defined as 70 to 180 mg/dL inclusive, and the average percent time in hypoglycemia (<70 and >54 mg/dL) and level 2 hypoglycemia (≤54 mg/dL). The TIR for a cohort of 20 virtual subjects of mGIPsim is 81.9% ± 7.4% (with no hypoglycemia or severe hypoglycemia) for the proposed MPC compared with 73.9% ± 7.6% (0.2% ± 0.1% in hypoglycemia and 0.1% ± 0.1% in level 2 hypoglycemia) for an MPC based on a recursive autoregressive exogenous (ARX) model.
Conclusions:
The adaptive MPC algorithm that incorporates prior knowledge in the recursive updating of the glucose prediction model can contribute to the development of fully automated artificial pancreas systems that can mitigate meal and physical activity disturbances.
Many fresh definitions of fractional derivatives have been suggested and used in recent years to produce mathematical models with memory, background, or non-local effects for a broad range of ...real-world structures. The primary aim of this article is to create and evaluate a fractional-order derivative for an extensive regulatory scheme for glucose-insulin regulation. The existence and uniqueness are determined by a fixed point theorem and an iterative scheme. We suggest an impulsive differential equation model study plasma glucose control for diabetic patients with impulsive insulin injections. It is regarded as a deterministic mathematical model related to the diabetes mellitus fractional derivatives. For fractional orders, numerical simulations are performed to demonstrate the impacts of varying the fractional-order to achieve the theoretical outcomes and comparison with the Caputo derivative are made. The results of these case studies indicate that this plasma glucose control of the fractional-order model is an appropriate candidate.
Open source artificial pancreas systems (OpenAPS) have gained considerable interest in the diabetes community. We analyzed continuous glucose monitoring (CGM) records of 80 OpenAPS users with type 1 ...diabetes (T1D). A total of 19 495 days (53.4 years) of CGM records were available. Mean glucose was 7.6 ± 1.1 mmol/L, time in range 3.9–10 mmol/L was 77.5 ± 10.5%, <3.9 mmol/L was 4.3 ± 3.6%, <3.0 mmol/L was 1.3 ± 1.9%, >10 mmol/L was 18.2 ± 11.0% and > 13.9 mmol/L was 4.1 ± 4.0%, respectively. In 34 OpenAPS users, additional CGM records were obtained while using sensor‐augmented pump therapy (SAP). After changing from SAP to OpenAPS, lower mean glucose (−0.6 ± 0.7; P < 0.0001), lower estimated HbA1c (−0.4 ± 0.5%; P < 0.0001), higher time in range 3.9–10 mmol/L (+9.3 ± 9.5%; P < 0.0001), less time < 3.0 mmol/L (−0.7 ± 2.2%; P = 0.0171), lower coefficient of variation (−2.4 ± 5.8; P = 0.0198) and lower mean of daily differences (−0.6 ± 0.9 mmol/L; P = 0.0005) was observed. Glycaemic control using OpenAPS was comparable with results of more rigorously developed and tested AP systems. However, OpenAPS was used by a highly selective, motivated and technology‐adept cohort, despite not being approved for the treatment of individuals with T1D.
•Reinforcement learning could be used to design a closed-loop artificial pancreas.•Reinforcement learning is a personalized solution to estimate insulin delivery.•Lack of focus on aspects that ...influence blood glucose level such as physical activity.•Need to perform clinical validation of the blood glucose control algorithms.•More frequent use of reinforcement learning for blood glucose control is foreseen.
Reinforcement learning (RL) is a computational approach to understanding and automating goal-directed learning and decision-making. It is designed for problems which include a learning agent interacting with its environment to achieve a goal. For example, blood glucose (BG) control in diabetes mellitus (DM), where the learning agent and its environment are the controller and the body of the patient respectively. RL algorithms could be used to design a fully closed-loop controller, providing a truly personalized insulin dosage regimen based exclusively on the patient’s own data.
In this review we aim to evaluate state-of-the-art RL approaches to designing BG control algorithms in DM patients, reporting successfully implemented RL algorithms in closed-loop, insulin infusion, decision support and personalized feedback in the context of DM.
An exhaustive literature search was performed using different online databases, analyzing the literature from 1990 to 2019. In a first stage, a set of selection criteria were established in order to select the most relevant papers according to the title, keywords and abstract. Research questions were established and answered in a second stage, using the information extracted from the articles selected during the preliminary selection.
The initial search using title, keywords, and abstracts resulted in a total of 404 articles. After removal of duplicates from the record, 347 articles remained. An independent analysis and screening of the records against our inclusion and exclusion criteria defined in Methods section resulted in removal of 296 articles, leaving 51 relevant articles. A full-text assessment was conducted on the remaining relevant articles, which resulted in 29 relevant articles that were critically analyzed. The inter-rater agreement was measured using Cohen Kappa test, and disagreements were resolved through discussion.
The advances in health technologies and mobile devices have facilitated the implementation of RL algorithms for optimal glycemic regulation in diabetes. However, there exists few articles in the literature focused on the application of these algorithms to the BG regulation problem. Moreover, such algorithms are designed for control tasks as BG adjustment and their use have increased recently in the diabetes research area, therefore we foresee RL algorithms will be used more frequently for BG control in the coming years. Furthermore, in the literature there is a lack of focus on aspects that influence BG level such as meal intakes and physical activity (PA), which should be included in the control problem. Finally, there exists a need to perform clinical validation of the algorithms.
Aims
To assess whether the dual‐hormone (insulin and glucagon) artificial pancreas reduces hypoglycaemia compared to the single‐hormone (insulin alone) artificial pancreas in outpatient settings ...during the day and night.
Material and methods
In a randomized, three‐way, crossover trial we compared the dual‐hormone artificial pancreas, the single‐hormone artificial pancreas and sensor‐augmented pump therapy (control) in 23 adults with type 1 diabetes. Each intervention was applied from 8 AM Day 1 to 8 PM Day 3 (60 hours) in outpatient free‐living conditions. The primary outcome was time spent with sensor glucose levels below 4.0 mmol/L. A P value of less than .017 was regarded as significant.
Results
The median difference between the dual‐hormone system and the single‐hormone system was −2.3% (P = .072) for time spent below 4.0 mmol/L, −1.3% (P = .017) for time below 3.5 mmol/L, and −0.7% (P = .031) for time below 3.3 mmol/L. Both systems reduced (P < .017) hypoglycaemia below 4.0, 3.5 and 3.3 mmol/L compared to control therapy, but reductions were larger with the dual‐hormone system than with the single‐hormone system (medians −4.0% vs −3.4% for 4.0 mmol/L; −2.7% vs −2.2% for 3.5 mmol/L; and −2.2% vs −1.2% for 3.3 mmol/L). There were 34 hypoglycaemic events (<3.0 mmol/L for 20 minutes) with control therapy, 14 with the single‐hormone system and 6 with the dual‐hormone system. These differences in hypoglycaemia were observed while mean glucose level was low and comparable in all interventions (P = NS).
Conclusions
The dual‐hormone artificial pancreas had the lowest risk of hypoglycaemia, but the differences were not statistically significant. Larger studies are needed.
Abstract The development of an artificial pancreas (AP) has been a topic of interest in the field of diabetes management for several decades. An AP system is designed to mimic the function of the ...pancreas by continuously monitoring blood glucose levels and delivering insulin or glucagon in response to changes in glucose concentration. Mathematical models play a crucial role in the development and evaluation of AP systems, as they enable the simulation and prediction of the system’s performance. This review paper provides an overview of the mathematical models used in AP research. The paper discusses the strengths and limitations of each type of model, as well as their applications in AP research. The review also highlights the challenges and opportunities in AP model development, such as the need for personalized models and the integration of data from multiple sources. Overall, this review provides a comprehensive understanding of the role of mathematical models in AP research and their potential for improving diabetes management.
The role of automated insulin delivery systems in diabetes is expanding. Hybrid closed-loop systems are being used in routine clinical practice for treating people with type 1 diabetes. ...Encouragingly, real-world data reflects the performance and usability observed in clinical trials. We review the commercially available hybrid closed-loop systems, their distinctive features and the associated real-world data. We also consider emerging indications for closed-loop systems, including the treatment of type 2 diabetes where variability of day-to-day insulin requirements is high, and other challenging applications for this technology. We discuss issues around access and implementation of closed-loop technology, and consider the limitations of present closed-loop systems, as well as innovative approaches that are being evaluated to improve their performance.
Objective
To describe predictors of hybrid closed loop (HCL) discontinuation and perceived barriers to use in youth with type 1 diabetes.
Subjects
Youth with type 1 diabetes (eligible age 2‐25 y; ...recruited age 8‐25 y) who initiated the Minimed 670G HCL system were followed prospectively for 6 mo in an observational study.
Research Design and Methods
Demographic, glycemic (time‐in‐range, HbA1c), and psychosocial variables Hypoglycemia Fear Survey (HFS); Problem Areas in Diabetes (PAID) were collected for all participants. Participants who discontinued HCL (<10% HCL use at clinical visit) completed a questionnaire on perceived barriers to HCL use.
Results
Ninety‐two youth (15.7 ± 3.6 y, HbA1c 8.8 ± 1.3%, 50% female) initiated HCL, and 28 (30%) discontinued HCL, with the majority (64%) discontinuing between 3 and 6 mo after HCL start. Baseline HbA1c predicted discontinuation (P = .026) with the odds of discontinuing 2.7 times higher (95% CI: 1.123, 6.283) for each 1% increase in baseline HbA1c. Youth who discontinued HCL rated difficulty with calibrations, number of alarms, and too much time needed to make the system work as the most problematic aspects of HCL. Qualitatively derived themes included technological difficulties (error alerts, not working correctly), too much work (calibrations, fingersticks), alarms, disappointment in glycemic control, and expense (cited by parents).
Conclusions
Youth with higher HbA1c are at greater risk for discontinuing HCL than youth with lower HbA1c, and should be the target of new interventions to support device use. The primary reasons for discontinuing HCL relate to the workload required to use HCL.
Objective: Recent years have seen an increase in machine learning (ML)-based blood glucose (BG) forecasting models, with a growing emphasis on potential application to hybrid or closed-loop ...predictive glucose controllers. However, current approaches focus on evaluating the accuracy of these models using benchmark data generated under the behavior policy, which may differ significantly from the data the model may encounter in a control setting. This study challenges the efficacy of such evaluation approaches, demonstrating that they can fail to accurately capture an ML-based model's true performance in closed-loop control settings. Methods: Forecast error measured using current evaluation approaches was compared to the control performance of two forecasters-a machine learning-based model (LSTM) and a rule-based model (Loop)- in silico when the forecasters were utilized with a model-based controller in a hybrid closed-loop setting. Results: Under current evaluation standards, LSTM achieves a significantly lower (better) forecast error than Loop with a root mean squared error (RMSE) of <inline-formula><tex-math notation="LaTeX">11.57 \pm 0.05 mg/dL</tex-math></inline-formula> vs. <inline-formula><tex-math notation="LaTeX">18.46 \pm 0.07 mg/dL</tex-math></inline-formula> at the 30-minute prediction horizon. Yet in a control setting, LSTM led to significantly worse control performance with only 77.14% (IQR 66.57-84.03) time-in-range compared to 86.20% (IQR 78.28-91.21) for Loop. Conclusion: Prevailing evaluation methods can fail to accurately capture the forecaster's performance when utilized in closed-loop settings. Significance: Our findings underscore the limitations of current evaluation standards and the need for alternative evaluation metrics and training strategies when developing BG forecasters for closed-loop control systems.