Dynamic Rule-Based Algorithm to Tune Insulin-on-Board Constraints for a Hybrid Artificial Pancreas System

Auto adaptation of closed-loop insulin delivery system using continuous reward functions and incremental discretization

Several closed or hybrid loop controllers for Blood Glucose (BG) regulation, which are also known as Artificial Pancreas (AP) Systems or Automated Insulin Delivery systems (AIDs), are in development worldwide. Most AIDs are designed and evaluated for short-term performance, with a particular emphasis on the post-meal period. However, if controllers are not adapted properly to account for variations in physiology that affect Insulin Sensitivity (IS), the AIDs may perform inadequately. In this work, the performance of two Reinforcement Learning (RL) agents trained under both piecewise and continuous reward functions is evaluated in-silico for long-term adaptation of a Fully Automated Insulin Delivery (fAID) system. An automatic adaptive discretization scheme that expands the state space as needed is also implemented to avoid disproportionate state space exploration. The proposed agents are evaluated for long-term adaptation of the Automatic Regulation of Glucose (ARG) algorithm, considering variations in IS. Results show that both RL agents have improved performance compared to a rule-based decision-making approach and the baseline controller for the majority of the adult population. Moreover, the use of a continuous shaped reward function proves to enhance the performance of the agents further than a piecewise one.

Safety and performance of a hybrid closed-loop insulin delivery system with carbohydrate suggestion in adults with type 1 diabetes prone to hypoglycemia

AIMS

To evaluate the safety and performance of a hybrid closed-loop (HCL) system with automatic carbohydrate suggestion in adults with type 1 diabetes (T1D) prone to hypoglycemia.

METHODS

A 32-hour in-hospital pilot study, including a night period, 4 meals and 2 vigorous unannounced 45-minute aerobic sessions, was conducted in 11 adults with T1D prone to hypoglycemia. The primary outcome was the percentage of time in range 70-180 mg/dL (TIR). Main secondary outcomes were time below range < 70 mg/dL (TBR < 70) and < 54 (TBR < 54). Data are presented as median (10th-90th percentile ranges).

RESULTS

The participants, 6 (54.5%) men, were 24 (22-48) years old, and had 22 (9-32) years of T1D duration. All of them regularly used an insulin pump and a continuous glucose monitoring system. The median TIR was 78.7% (75.6-91.2): 92.7% (68.2-100.0) during exercise and recovery period, 79.3% (34.9-100.0) during postprandial period, and 95.4% (66.4-100.0) during overnight period. The TBR < 70 and TBR < 54 were 0.0% (0.0-6.6) and 0.0% (0.0-1.2), respectively. A total of 4 (3-9) 15-g carbohydrate suggestions were administered per person. No severe acute complications occurred during the study.

CONCLUSIONS

The HCL system with automatic carbohydrate suggestion performed well and was safe in this population during challenging conditions in a hospital setting.

Using Continuous Glucose Monitoring Values for Bolus Size Calculation in Smart Multiple Daily Injection Systems: No Negative Impact on Post-bolus Glycemic Outcomes Found in Real-World Data

BACKGROUND

Recent evidence shows that it may be safe to estimate bolus sizes based on continuous glucose monitoring (CGM) rather than blood glucose (BG) values using glycemic trend-adjusted bolus calculators. Users may already be doing this in the real world, though it is unclear whether this is safe or effective for calculators not employing trend adjustment.

METHODS

We assessed real-world data from a smart multiple daily injections (MDIs) device users with a CGM system, hypothesizing that four-hour post-bolus outcomes using CGM values are not inferior to those using BG values. Our data set included 184 users and spanned 18 months with 79 000 bolus observations. We tested differences using logistic regression predicting CGM or BG value usage based on outcomes and confirmed initial results using a mixed model regression accounting for within-subject correlations.

RESULTS

Comparing four-hour outcomes for bolus events using CGM and BG values revealed no differences using our initial approach (P > .183). This finding was confirmed by our mixed model regression approach in all cases (P > .199), except for times below range outcomes. Higher times below range were predictive of lower odds of CGM-based bolus calculations (OR = 0.987, P < .0001 and OR = 0.987, P = .0276, for time below 70 and 54 mg/dL, respectively).

CONCLUSIONS

We found no differences in four-hour post-bolus glycemic outcomes when using CGM or BG except for time below range, which showed evidence of being lower for CGM. Though preliminary, our results confirm prior findings showing non-inferiority of using CGM values for bolus calculation compared with BG usage in the real world.

An Intelligent Model-Based Effective Approach for Glycemic Control in Type-1 Diabetes

Type-1 diabetes mellitus (T1DM) is a challenging disorder which essentially involves regulation of the glucose levels to avoid hyperglycemia as well as hypoglycemia. For this purpose, this research paper proposes and develops control algorithms using an intelligent predictive control model, which is based on a UVA/Padova metabolic simulator. The primary objective of the designed control laws is to provide an automatic blood glucose control in insulin-dependent patients so as to improve their life quality and to reduce the need of an extremely demanding self-management plan. Various linear and nonlinear control algorithms have been explored and implemented on the estimated model. Linear techniques include the Proportional Integral Derivative (PID) and Linear Quadratic Regulator (LQR), and nonlinear control strategy includes the Sliding Mode Control (SMC), which are implemented in this research work for continuous monitoring of glucose levels. Performance comparison based on simulation results demonstrated that SMC proved to be most efficient in terms of regulating glucose profile to a reference level of 70 mg/dL compared to the classical linear techniques. A brief comparison is presented between the linear techniques (PID and LQR), and nonlinear technique (SMC) for analysis purposes proving the efficacy of the design.

A simulator with realistic and challenging scenarios for virtual T1D patients undergoing CSII and MDI therapy

In silico simulations have become essential for the development of diabetes treatments. However, currently available simulators are not challenging enough and often suffer from limitations in insulin and meal absorption variability, which is unable to realistically reflect the dynamics of people with type 1 diabetes (T1D). Additionally, T1D simulators are mainly designed for the testing of continuous subcutaneous insulin infusion (CSII) therapies. In this work, a simulator is presented that includes a generated virtual patient (VP) cohort and both fast- and long-acting Glargine-100 U/ml (Gla-100), Glargine-300 U/ml (Gla-300), and Degludec-100 U/ml (Deg-100) insulin models. Therefore, in addition to CSII therapies, multiple daily injections (MDI) therapies can also be tested. The Hovorka model and its published parameter probability distributions were used to generate cohorts of VPs that represent a T1D population. Valid patients are filtered through restrictions that guarantee that they are physiologically acceptable. To obtain more realistic scenarios, basal insulin profile patterns from the literature have been used to identify variability in insulin sensitivity. A library of mixed meals identified from real data has also been included. This work presents and validates a methodology for the creation of realistic VP cohorts that include physiological variability and a simulator that includes challenging and realistic scenarios for in silico testing. A cohort of 47 VPs has been generated and in silico simulations of both CSII and MDI therapies were performed in open-loop. The simulation outcome metrics were contrasted with literature results.

Practical constraint definition in safety schemes for artificial pancreas systems

Introduction:

Artificial pancreas systems usually define an insulin-on-board constraint ([Formula: see text]) for safety schemes to limit the insulin infusion and avoid hypoglycemia during the closed-loop performance. Several methods have been proposed with impractical considerations requiring information from the prandial events or complex procedures for ambulatory use.

Methods:

This paper presents a simple method that consists of two novel rules that allow finding an [Formula: see text] based only on common clinical parameters that do not require patient intervention. The method robustness was evaluated using a control system coupled to a safety layer under demanding scenarios implemented on the FDA-approved simulator for preclinical studies.

Results:

The method maintains a safe performance, even in the face of interpatient variability, hybrid and fully automatic implementations of an artificial pancreas system, and uncertain settings. Both proposed rules work as effectively or even better and without the patient intervention than other methods that have already been clinically validated.

Conclusion:

This method can be used to define a constant [Formula: see text] that ensures performance and safety of the control system, even under scenarios with incorrect clinical data. Unlike other methods, this method only requires reliable information that is easily obtained from the patient, such as their total daily dose of insulin or body mass.

Fault Tolerant Strategies for Automated Insulin Delivery Considering the Human Component: Current and Future Perspectives

Automated Insulin Delivery (AID) are systems developed for daily use by people with type 1 diabetes (T1D). To ensure the safety of users, it is essential to consider how the human factor affects the performance and safety of these devices. While there are numerous publications on hardware-related failures of AID systems, there are few studies on the human component of the system. From a control point of view, people with T1D using AID systems are at the same time the plant to be controlled and the plant operator. Therefore, users may induce faults in the controller, sensors, actuators, and the plant itself. Strategies to cope with the human interaction in AID systems are needed for further development of the technology. In this paper, we present an analysis of potential faults introduced by AID users when the system is under normal operation. This is followed by a review of current fault tolerant control (FTC) approaches to identify missing areas of research. The paper concludes with a discussion on future directions for the new generation of FTC AID systems.

A Hybrid Automata Approach for Monitoring the Patient in the Loop in Artificial Pancreas Systems

The use of automated insulin delivery systems has become a reality for people with type 1 diabetes (T1D), with several hybrid systems already on the market. One of the particularities of this technology is that the patient is in the loop. People with T1D are the plant to control and also a plant operator, because they may have to provide information to the control loop. The most immediate information provided by patients that affects performance and safety are the announcement of meals and exercise. Therefore, to ensure safety and performance, the human factor impact needs to be addressed by designing fault monitoring strategies. In this paper, a monitoring system is developed to diagnose potential patient modes and faults. The monitoring system is based on the residual generation of a bank of observers. To that aim, a linear parameter varying (LPV) polytopic representation of the system is adopted and a bank of Kalman filters is designed using linear matrix inequalities (LMI). The system uncertainty is propagated using a zonotopic-set representation, which allows determining confidence bounds for each of the observer outputs and residuals. For the detection of modes, a hybrid automaton model is generated and diagnosis is performed by interpreting the events and transitions within the automaton. The developed system is tested in simulation, showing the potential benefits of using the proposed approach for artificial pancreas systems.

Artificial Pancreas With Carbohydrate Suggestion Performance for Unannounced and Announced Exercise in Type 1 Diabetes

OBJECTIVE

To evaluate the safety and performance of a new multivariable closed-loop (MCL) glucose controller with automatic carbohydrate recommendation during and after unannounced and announced exercise in adults with type 1 diabetes (T1D).

RESEARCH DESIGN AND METHODS

A randomized, 3-arm, crossover clinical trial was conducted. Participants completed a heavy aerobic exercise session including three 15-minute sets on a cycle ergometer with 5 minutes rest in between. In a randomly determined order, we compared MCL control with unannounced (CLNA) and announced (CLA) exercise to open-loop therapy (OL). Adults with T1D, insulin pump users, and those with hemoglobin (Hb)A1c between 6.0% and 8.5% were eligible. We investigated glucose control during and 3 hours after exercise.

RESULTS

Ten participants (aged 40.8 ± 7.0 years; HbA1c of 7.3 ± 0.8%) participated. The use of the MCL in both closed-loop arms decreased the time spent <70 mg/dL of sensor glucose (0.0%, [0.0-16.8] and 0.0%, [0.0-19.2] vs 16.2%, [0.0-26.0], (%, [percentile 10-90]) CLNA and CLA vs OL respectively; P = 0.047, P = 0.063) and the number of hypoglycemic events when compared with OL (CLNA 4 and CLA 3 vs OL 8; P = 0.218, P = 0.250). The use of the MCL system increased the proportion of time within 70 to 180 mg/dL (87.8%, [51.1-100] and 91.9%, [58.7-100] vs 81.1%, [65.4-87.0], (%, [percentile 10-90]) CLNA and CLA vs OL respectively; P = 0.227, P = 0.039). This was achieved with the administration of similar doses of insulin and a reduced amount of carbohydrates.

CONCLUSIONS

The MCL with automatic carbohydrate recommendation performed well and was safe during and after both unannounced and announced exercise, maintaining glucose mostly within the target range and reducing the risk of hypoglycemia despite a reduced amount of carbohydrate intake.Register Clinicaltrials.gov: NCT03577158.

An Insulin Bolus Advisor for Type 1 Diabetes Using Deep Reinforcement Learning

(1) Background: People living with type 1 diabetes (T1D) require self-management to maintain blood glucose (BG) levels in a therapeutic range through the delivery of exogenous insulin. However, due to the various variability, uncertainty and complex glucose dynamics, optimizing the doses of insulin delivery to minimize the risk of hyperglycemia and hypoglycemia is still an open problem. (2) Methods: In this work, we propose a novel insulin bolus advisor which uses deep reinforcement learning (DRL) and continuous glucose monitoring to optimize insulin dosing at mealtime. In particular, an actor-critic model based on deep deterministic policy gradient is designed to compute mealtime insulin doses. The proposed system architecture uses a two-step learning framework, in which a population model is first obtained and then personalized by subject-specific data. Prioritized memory replay is adopted to accelerate the training process in clinical practice. To validate the algorithm, we employ a customized version of the FDA-accepted UVA/Padova T1D simulator to perform in silico trials on 10 adult subjects and 10 adolescent subjects. (3) Results: Compared to a standard bolus calculator as the baseline, the DRL insulin bolus advisor significantly improved the average percentage time in target range (70-180 mg/dL) from 74.1%±8.4% to 80.9%±6.9% (p<0.01) and 54.9%±12.4% to 61.6%±14.1% (p<0.01) in the the adult and adolescent cohorts, respectively, while reducing hypoglycemia. (4) Conclusions: The proposed algorithm has the potential to improve mealtime bolus insulin delivery in people with T1D and is a feasible candidate for future clinical validation.