Search

The surveillance error grid (SEG) analysis is a tool for analysis and visualization of blood glucose monitoring (BGM) errors, based on the opinions of 206 diabetes clinicians who rated 4 distinct treatment scenarios. Resulting from this large-scale inquiry is a matrix of 337 561 risk ratings, 1 for each pair of (reference, BGM) readings ranging from 20 to 580 mg/dl. The computation of the SEG is therefore complex and in need of automation. The SEG software introduced in this article automates the task of assigning a degree of risk to each data point for a set of measured and reference blood glucose values so that the data can be distributed into 8 risk zones. The software's 2 main purposes are to (1) distribute a set of BG Monitor data into 8 risk zones ranging from none to extreme and (2) present the data in a color coded display to promote visualization. Besides aggregating the data into 8 zones corresponding to levels of risk, the SEG computes the number and percentage of data pairs in each zone and the number/percentage of data pairs above/below the diagonal line in each zone, which are associated with BGM errors creating risks for hypo- or hyperglycemia, respectively. To illustrate the action of the SEG software we first present computer-simulated data stratified along error levels defined by ISO 15197:2013. This allows the SEG to be linked to this established standard. Further illustration of the SEG procedure is done with a series of previously published data, which reflect the performance of BGM devices and test strips under various environmental conditions. We conclude that the SEG software is a useful addition to the SEG analysis presented in this journal, developed to assess the magnitude of clinical risk from analytically inaccurate data in a variety of high-impact situations such as intensive care and disaster settings., (© 2014 Diabetes Technology Society.)

Objective: We estimate the effect size of hypoglycemia risk reduction on closed-loop control (CLC) versus open-loop (OL) sensor-augmented insulin pump therapy in supervised outpatient setting., Research Design and Methods: Twenty patients with type 1 diabetes initiated the study at the Universities of Virginia, Padova, and Montpellier and Sansum Diabetes Research Institute; 18 completed the entire protocol. Each patient participated in two 40-h outpatient sessions, CLC versus OL, in randomized order. Sensor (Dexcom G4) and insulin pump (Tandem t:slim) were connected to Diabetes Assistant (DiAs)-a smartphone artificial pancreas platform. The patient operated the system through the DiAs user interface during both CLC and OL; study personnel supervised on site and monitored DiAs remotely. There were no dietary restrictions; 45-min walks in town and restaurant dinners were included in both CLC and OL; alcohol was permitted., Results: The primary outcome-reduction in risk for hypoglycemia as measured by the low blood glucose (BG) index (LGBI)-resulted in an effect size of 0.64, P = 0.003, with a twofold reduction of hypoglycemia requiring carbohydrate treatment: 1.2 vs. 2.4 episodes/session on CLC versus OL (P = 0.02). This was accompanied by a slight decrease in percentage of time in the target range of 3.9-10 mmol/L (66.1 vs. 70.7%) and increase in mean BG (8.9 vs. 8.4 mmol/L; P = 0.04) on CLC versus OL., Conclusions: CLC running on a smartphone (DiAs) in outpatient conditions reduced hypoglycemia and hypoglycemia treatments when compared with sensor-augmented pump therapy. This was accompanied by marginal increase in average glycemia resulting from a possible overemphasis on hypoglycemia safety., (© 2014 by the American Diabetes Association.)

The Hypoglycemia-Hyperglycemia Minimizer (HHM) System aims to mitigate glucose excursions by preemptively modulating insulin delivery based on continuous glucose monitor (CGM) measurements. The "aggressiveness factor" is a key parameter in the HHM System algorithm, affecting how readily the system adjusts insulin infusion in response to changing CGM levels. Twenty adults with type 1 diabetes were studied in closed-loop in a clinical research center for approximately 26 hours. This analysis focused on the effect of the aggressiveness factor on the insulin dosing characteristics of the algorithm and, to a lesser extent, on the glucose control results observed. As the aggressiveness factor increased from conservative to medium to aggressive: the maximum observed insulin dose delivered by the algorithm—which is designed to give doses that are corrective in nature every 5 minutes—increased (1.00 vs 1.15 vs 2.20 U, respectively); tendency to adhere to the subject's nominal basal dose decreased (61.9% vs 56.6% vs 53.4%); and readiness to decrease insulin below basal also increased (18.4% vs 19.4% vs 25.2%). Glucose analyses by both CGM and Yellow Springs Instruments (YSI) indicated that the aggressive setting of the algorithm resulted in the least time spent at levels >180 mg/dL, and the most time spent between 70-180 mg/dL. There was no severe hyperglycemia, diabetic ketoacidosis, or severe hypoglycemia for any of the aggressiveness values investigated. These analyses underscore the importance of investigating the sensitivity of the HHM System to its key parameters, such as the aggressiveness factor, to guide future development decisions., (© 2014 Diabetes Technology Society.)

Background: Laboratory hemoglobin A1c (HbA1c) assays are typically done only every few months. However, self-monitored blood glucose (SMBG) readings offer the possibility for real-time estimation of HbA1c. We present a new dynamical method tracking changes in average glycemia to provide real-time estimation of A1c (eA1c)., Materials and Methods: A new two-step algorithm was constructed that includes: (1) tracking fasting glycemia to compute base eA1c updated with every fasting SMBG data point and (2) calibration of the base eA1c trace with monthly seven-point SMBG profiles to capture the principal components of blood glucose variability and produce eA1c. A training data set (n=379 subjects) was used to estimate model parameters. The model was then fixed and applied to an independent test data set (n=375 subjects). Accuracy was evaluated in the test data set by computing mean absolute deviation (MAD) and mean absolute relative deviation (MARD) of eA1c from reference HbA1c, as well as eA1c-HbA1c correlation., Results: MAD was 0.50, MARD was 6.7%, and correlation between eA1c and reference HbA1c was r=0.76. Using an HbA1c error grid plot, 77.5% of all eA1c fell within 10% from reference HbA1c, and 97.9% fell within 20% from reference., Conclusions: A dynamical estimation model was developed that achieved accurate tracking of average glycemia over time. The model is capable of working with infrequent SMBG data typical for type 2 diabetes, thereby providing a new tool for HbA1c estimation at the patient level. The computational demands of the procedure are low; thus it is readily implementable into home SMBG meters. Real-time HbA1c estimation could increase patients' motivation to improve diabetes control.

Background: Insulin-induced hypoglycemia is as a critical barrier in the treatment of type 1 diabetes mellitus patients and may lead to unconsciousness, brain damage, or even death. Clinically, glucagon is used as a rescue drug to treat severe hypoglycemic episodes. More recently, in a bihormonal closed-loop glucose control, glucagon has been used subcutaneously along with insulin for protection against hypoglycemia. In this context, small doses of glucagon are frequently administered. The efficacy and safety of such systems, however, require precise information on the pharmacokinetics of the glucagon transport from the administrative site to the circulation, which is currently lacking. The goal of this work is to address this need by developing and validating a mathematical model of exogenous glucagon transport to the plasma., Materials and Methods: Eight pharmacokinetic models with various levels of complexity were fitted to nine clinical datasets. An optimal model was chosen in two consecutive steps. At Step 1, all models were screened for parameter identifiability (discarding the unidentifiable candidates). At Step 2, the remaining models are compared based on Bayesian information criterion., Results: At Step 1, two models were removed for higher parameter fractional SDs. Another three were discarded for location of their optimal parameters on the parameter search boundaries. At Step 2, an optimal model was selected based on the Bayesian information criterion. It has a simple linear structure, assuming that glucagon is injected into one compartment, from where it enters a pool for a slower release into a third, plasma compartment. In the first and third compartments, glucagon is cleared at a rate proportional to its concentration., Conclusions: A linear kinetic model of glucagon intervention has been developed and validated. It is expected to provide guidance for glucagon delivery and the construction of preclinical simulation testing platforms.

Objective: To evaluate the feasibility of a wearable artificial pancreas system, the Diabetes Assistant (DiAs), which uses a smart phone as a closed-loop control platform., Research Design and Methods: Twenty patients with type 1 diabetes were enrolled at the Universities of Padova, Montpellier, and Virginia and at Sansum Diabetes Research Institute. Each trial continued for 42 h. The United States studies were conducted entirely in outpatient setting (e.g., hotel or guest house); studies in Italy and France were hybrid hospital-hotel admissions. A continuous glucose monitoring/pump system (Dexcom Seven Plus/Omnipod) was placed on the subject and was connected to DiAs. The patient operated the system via the DiAs user interface in open-loop mode (first 14 h of study), switching to closed-loop for the remaining 28 h. Study personnel monitored remotely via 3G or WiFi connection to DiAs and were available on site for assistance., Results: The total duration of proper system communication functioning was 807.5 h (274 h in open-loop and 533.5 h in closed-loop), which represented 97.7% of the total possible time from admission to discharge. This exceeded the predetermined primary end point of 80% system functionality., Conclusions: This study demonstrated that a contemporary smart phone is capable of running outpatient closed-loop control and introduced a prototype system (DiAs) for further investigation. Following this proof of concept, future steps should include equipping insulin pumps and sensors with wireless capabilities, as well as studies focusing on control efficacy and patient-oriented clinical outcomes.

Safety measures to prevent or mitigate hypoglycemia are an important component of open loop, closed loop, and advisory mode insulin therapy control settings in type 1 diabetes. In recent work, we introduce a method for the automatic, gradual attenuation of the insulin pump delivery rate when a risk of hypoglycemia is detected, a method that we refer to as brakes. In the methods presented here, we demonstrate the use of historical glucose measurement data to inform and enhance the ability of the brakes to prevent hypoglycemia in real-time. The updated brakes are based on a patient-specific, time-varying model that reflects the typical trajectory of glycemic fluctuations throughout the day. Historical heightened risk of hypoglycemia throughout the day prompts an increase in the aggressiveness of insulin attenuation as compared to the original brakes that are based on real-time data alone. Through the use of available real-time data supplemented with historical glucose information to assess hypoglycemic risk, we are able to better anticipate and prevent hypoglycemia., (Copyright © 2012 Elsevier Ireland Ltd. All rights reserved.)

Aims/hypothesis: Insulin delivery to muscle is rate-limiting for insulin's metabolic action and is regulated by insulin's own action to increase skeletal muscle blood flow and to recruit microvasculature. Microvascular dysfunction has been observed in insulin resistant states. We investigated the relation between insulin's action to recruit microvasculature and its metabolic action in type 1 diabetes., Methods: Near euglycaemia was obtained by an overnight insulin infusion during 17 inpatient admissions of participants with type 1 diabetes. This was followed by a 2 h 1 mU kg⁻¹ min⁻¹ euglycaemic-hyperinsulinaemic clamp. Microvascular blood volume (MBV) was assessed using contrast-enhanced ultrasound 10 min before and 30 min after starting the clamp., Results: We observed that, after overnight modest hyperinsulinaemia (average ≈ 286 pmol/l), MBV was positively related to the steady-state insulin sensitivity measured during the subsequent clamp (r = 0.62, p = 0.008). The more marked hyperinsulinaemia during the clamp (average steady-state insulin ≈ 900 pmol/l) increased MBV in the more insulin resistant participants within 30 min but not in the insulin sensitive participants. The change in MBV during the clamp was negatively correlated to the insulin sensitivity (r = -0.55, p = 0.022). As a result, MBV after 30 min of marked hyperinsulinaemia was comparable between the insulin sensitive and resistant participants., Conclusions/interpretation: We conclude that moderate overnight hyperinsulinaemia recruited microvasculature in the more sensitive participants, while higher levels of plasma insulin were needed for more insulin resistant participants. This suggests that microvascular responsiveness to insulin is one determinant of metabolic insulin sensitivity in type 1 diabetes.

Glucagon counterregulation (GCR) protects against hypoglycemia, but is impaired in type 1 diabetes (T1DM). A model-based analysis of in vivo animal data predicts that the GCR defects are linked to basal hyperglucagonemia. To test this hypothesis we studied the relationship between basal glucagon (BasG) and the GCR response to hypoglycemia in 29 hyperinsulinemic clamps in T1DM patients. Glucose levels were stabilized in euglycemia and then steadily lowered to 50 mg/dL. Glucagon was measured before induction of hypoglycemia and at 10 min intervals after glucose reached levels below 70 mg/dL. GCR was assessed by CumG, the cumulative glucagon levels above basal; MaxG, the maximum glucagon response; and RIG, the relative increase in glucagon over basal. Analysis of the results was performed with our mathematical model of GCR. The model describes interactions between islet peptides and glucose, reproduces the normal GCR axis and its impairment in diabetes. It was used to identify a control mechanism consistent with the observed link between BasG and GCR. Analysis of the clinical data showed that higher BasG was associated with lower GCR response. In particular, CumG and RIG correlated negatively with BasG (r = -0.46, p = 0.012 and r = -0.74, p < 0.0001 respectively) and MaxG increased linearly with BasG at a rate less than unity (p < 0.001). Consistent with these results was a model of GCR in which the secretion of glucagon has two components. The first is under (auto) feedback control and drives a pulsatile GCR and the second is feedback independent (basal secretion) and its increase suppresses the GCR. Our simulations showed that this model explains the observed relationships between BasG and GCR during a three-fold simulated increase in BasG. Our findings support the hypothesis that basal hyperglucagonemia contributes to the GCR impairment in T1DM and show that the predictive power of our GCR animal model applies to human pathophysiology in T1DM.

Contrast-enhanced ultrasound (CEU) is an ultrasound imaging technique used to assess tissue perfusion. Analysis of microvascular recruitment necessitates the definition of a region of interest (ROI) containing exclusively the tissues to be studied. Conventional ROI selection requires examining the images and drawing the ROI by hand, making the analysis of CEU images non-reproducible and analyst-dependent. We have designed a systematic ROI selection method that is both reproducible and analyst-independent. Microvascular blood volume (MBV) assessed in 21 sequences of images was used to correlate the systematic ROI selection method with the conventional method performed by two independent analysts (correlation of 0.88 and 0.87 respectively) and the MBV sample distribution from the systematic method was not significantly different from those obtained from the conventional one. Using the systematic method, we found no significant insulin-induced capillary recruitment in subjects with type 1 diabetes mellitus, which might be related to the observed low glucose uptake during the hyperinsulinemic euglycemic clamp compared to healthy patients., (Copyright © 2010 Elsevier Ireland Ltd. All rights reserved.)

Background: Hypoglycemia has been identified as a primary barrier to optimal management of diabetes. This observation, in conjunction with the introduction of continuous glucose monitoring (CGM) devices, has set the stage for achieving tight glycemic control with systems that adjust the insulin pump settings based on measured glucose concentrations. Because system safety and system reliability are key considerations, there is a need for algorithms that reduce the risk of hypoglycemia in closed-loop, open-loop, and advisory-mode systems. More specifically, the algorithm presented here is formulated as a component of the independent safety system module proposed in the modular control-to-range architecture., Methods: We developed two algorithms for attenuating insulin pump injections, which we refer to as Brakes and Power Brakes: Brakes is a pump attenuation function computed using CGM information only, while Power Brakes is an attenuation function in which a metabolic state observer with insulin input is used in addition to CGM information to inform the level of pump attenuation. These algorithms modulate the insulin pump delivery so that the insulin injection rate is dramatically reduced when the risk of hypoglycemia is high. Additionally, we combined these algorithms with an alert system that indicates a level of hypoglycemic risk to the user., Results: We demonstrated the effectiveness of Brakes and Power Brakes in reducing the incidence of hypoglycemia in two simulated scenarios: an elevated basal rate scenario and a scenario in which a bolus is delivered for a meal that is skipped. For these scenarios, the incidence of hypoglycemia using Power Brakes was reduced by 88 and 94%, respectively, where we defined hypoglycemia based on the American Diabetes Association guidelines for defining and reporting as 70 mg/dl. In the elevated basal rate scenario, no rebounds above 180 mg/dl (the desired upper limit of the control-to-range protocol) following hypoglycemia were shown to occur. We demonstrated the way the hypoglycemia alert system can trigger the intake of carbohydrates to reduce the incidence of hypoglycemia by 98%., Conclusions: This article offers, for the first time, a method for smoothly reducing insulin delivery rate to prevent hypoglycemia in individuals with type 1 diabetes mellitus based on a mathematically formal assessment of hypoglycemic risk. In silico, we demonstrate the way this method can prevent hypoglycemia while avoiding hyperglycemia rebounds that exceed 180 mg/dl. In conjunction with the pump attenuation algorithms, this article also proposes a system for alerting an individual of their hypoglycemic risk that contains three "levels" of alerts in the form of a traffic light. This alert system can be used in an advisory mode setting to alert the user to take action when hypoglycemia is imminent ("red" light) or in a closed-loop setting where initiation of rescue action begins when the red light alert is triggered., (© 2010 Diabetes Technology Society.)

Background: For insulin therapy to successfully maintain blood glucose (BG) levels of patients with type 1 diabetes mellitus (T1DM) in normoglycemia, it is necessary to understand if the metabolic effect of insulin across the BG range is linear or not., Methods: We assess the ability of insulin to lower BG in patients with T1DM in hypoglycemia and hyperglycemia. The net metabolic effect of insulin, defined as the total effect resulting from both reduced endogenous glucose production and increased glucose uptake, was used to define the insulin effectiveness (IE), a measure that indicates the amplitude of glucose lowering that a unit of active insulin can achieve at a given BG level. The IE was assessed in hypoglycemia and hyperglycemia through two separate studies. In the first study, patients were subjected to a hyperinsulinemic euglycemic and hypoglycemic glucose clamp. In the second study, another group of patients were clamped at a hyperglycemic level., Results: The IE increased by 75% when BG dropped from 90 to 50 mg/dl at a steady rate of 1 mg/dl/min and decreased by 10% when BG was increased from 100 to 200 mg/dl., Conclusions: The net metabolic effect of insulin is nonlinear across the BG range and is amplified in hypoglycemia and dampened in hyperglycemia. Most importantly, the BG lowering per unit of insulin is accelerated when falling into hypoglycemia. The understanding of the accelerated risk for hypoglycemia with falling glucose levels will help the design of more robust hypoglycemia prevention and detection systems., (2010 Diabetes Technology Society.)

Background: Clinical trials assessing the impact of errors in self-monitoring of blood glucose (SMBG) on the quality of glycemic control in diabetes are inherently difficult to execute. Consequently, the objectives of this study were to employ realistic computer simulation based on a validated model of the human metabolic system and to provide potentially valuable information about the relationships among SMBG errors, risk for hypoglycemia, glucose variability, and long-term glycemic control., Methods: Sixteen thousand computer simulation trials were conducted using 100 simulated adults with type 1 diabetes. Each simulated subject was used in four simulation experiments aiming to assess the impact of SMBG errors on detection of hypoglycemia (experiment 1), risk for hypoglycemia (experiment 2), glucose variability (experiment 3), and long-term average glucose control, i.e., estimated hemoglobin A1c (HbA1c)(experiment 4). Each experiment was repeated 10 times at each of four increasing levels of SMBG errors: 5, 10, 15, and 20% deviation from the true blood glucose value., Results: When the permitted SMBG error increased from 0 to 5-10% to 15-20%-the current level allowed by International Organization for Standardization 15197-(1) the probability for missing blood glucose readings of 60 mg/dl increased from 0 to 0-1% to 3.5-10%; (2) the incidence of hypoglycemia, defined as reference blood glucose

Introduction: A simulation model of a glucose-insulin system accounting for physical activity is needed to reliably simulate normal life conditions, thus accelerating the development of an artificial pancreas. In fact, exercise causes a transient increase of insulin action and may lead to hypoglycemia. However, physical activity is difficult to model. In the past, it was described indirectly as a rise in insulin. Recently, a new parsimonious model of exercise effect on glucose homeostasis has been proposed that links the change in insulin action and glucose effectiveness to heart rate (HR). The aim of this study was to plug this exercise model into our recently proposed large-scale simulation model of glucose metabolism in type 1 diabetes to better describe normal life conditions., Methods: The exercise model describes changes in glucose-insulin dynamics in two phases: a rapid on-and-off change in insulin-independent glucose clearance and a rapid-on/slow-off change in insulin sensitivity. Three candidate models of glucose effectiveness and insulin sensitivity as a function of HR have been considered, both during exercise and recovery after exercise. By incorporating these three models into the type 1 diabetes model, we simulated different levels (from mild to moderate) and duration of exercise (15 and 30 minutes), both in steady-state (e.g., during euglycemic-hyperinsulinemic clamp) and in nonsteady state (e.g., after a meal) conditions., Results: One candidate exercise model was selected as the most reliable., Conclusions: A type 1 diabetes model also describing physical activity is proposed. The model represents a step forward to accurately describe glucose homeostasis in normal life conditions; however, further studies are needed to validate it against data., (© Diabetes Technology Society)

Background: Most insulin pumps used for the treatment of diabetes perform subcutaneous insulin injections by pulses. The purpose of this work is to analyze the effects of pulsatile injections of modern insulins on plasma insulin levels compared with a continuous insulin infusion., Method: We simulate pulsatile implementations of a basal rate profile over a day on a type 1 diabetes mellitus patient using insulin lispro. Pulse periods were varied between 1 and 60 min, and random pump errors were included, modeled as white noise, 1/f noise, or 1/f(2) noise with relative standard deviations up to 10% of the pump output., Results: Oscillations in plasma insulin caused by the pulsatile injections were not significant with respect to the global variations for pulse periods below 15 min. This cutoff period was found to be robust to random pump errors with standard deviations up to 10% of the pump output and hence solely determined by the insulin kinetics. Additionally, we showed that the pulse period achieving the best implementation of a continuous profile is an increasing function of the error variance for a given type of noise., Conclusions: Our findings support that continuous insulin infusion can be implemented by a pulsatile injection of insulin as infrequent as a pulse every 15 min without significant effects on plasma insulin levels. If clinically confirmed, this result would have important consequences on the design and in silico testing of automated insulin treatment strategies, as increased delivery intervals imply higher accuracy of insulin delivery and facilitated implementations of closed-loop control algorithms.

Background: The objective of this article was to focus on the application of harmonic decomposition to continuous glucose monitor (CGM) measurements. We show evidence of an attenuation of fast variations of interstitial glucose when compared to blood in type 1 diabetes mellitus (T1DM) and, using information theory, propose optimal sampling schedules associated with the use and study of CGMs., Method: Using a cohort of 26 T1DM subjects, wearing two Navigator sensors for 1 to 3 days, we analyzed the frequency content of each glucose signal and derived across subject frequency cutoffs using discrete Fourier transform and common signal processing techniques., Results: We observed a significant difference in the frequency content of blood glucose compared to interstitial glucose in T1DM, providing evidence toward the existence of a diffusion process between blood and interstitial glucose, acting as a low-pass filter. Furthermore, we obtained a 15-minutes sampling schedule for optimal comparison of CGM values to blood reference., Conclusion: Blood glucose and interstitial glucose have different dynamics, as shown by harmonic analysis, and these differences have consequences on advisable schedules for accuracy studies of CGMS.

This article presents a mathematical model of glucose homeostasis that is valid during physical activity. Known changes in glucose dynamics during exercise were accounted for in the model, and exercise itself was detected and quantified through heart rate (beats per minute). The model was successfully fit to 21 type 1 diabetic subjects during a hyperinsulemic clamp protocol, and performance of the new model was compared with the standard minimal model of glucose kinetics that it was derived from.

Background: We investigated the applicability of linear quadratic Gaussian (LQG) methodology to the subcutaneous blood glucose regulation problem. We designed an LQG-based feedback control algorithm using linearization of a previously published metabolic model of type 1 diabetes. A key feature of the controller is a Kalman filter used to estimate metabolic states of the patient based on continuous glucose monitoring. Insulin infusion is computed from linear quadratic regulator feedback gains applied to these estimates, generally seeking to minimize squared deviations from a target glucose concentration and basal insulin rate. We evaluated in silico subject-specific LQG control and compared it to preexisting proportional-integral-derivative control.

In this paper we present a mathematical model and a computer simulation of the relationship between behavioral control of Type 1 diabetes (T1DM), blood glucose (BG) variability, and hypoglycemia-associated autonomic failure (HAAF). A stochastic behavioral self-control process was coupled with a dynamical system simulation of the dampening effect of counterregulation on BG oscillations. The resulting bio-behavioral control system was able to reproduce characteristics of hypoglycemic events in the field, such as temporal clustering of hypoglycemic episodes associated with HAAF, and occurrence of severe hypoglycemia as a result of periods of HAAF augmented by increased BG variability.