Jorge Bondia

Estimating Plasma Glucose from Interstitial Glucose: The Issue of Calibration Algorithms in Commercial Continuous Glucose Monitoring Devices

Evaluation of metabolic control of diabetic people has been classically performed measuring glucose concentrations in blood samples. Due to the potential improvement it offers in diabetes care, continuous glucose monitoring (CGM) in the subcutaneous tissue is gaining popularity among both patients and physicians. However, devices for CGM measure glucose concentration in compartments other than blood, usually the interstitial space. This means that CGM need calibration against blood glucose values, and the accuracy of the estimation of blood glucose will also depend on the calibration algorithm. The complexity of the relationship between glucose dynamics in blood and the interstitial space, contrasts with the simplistic approach of calibration algorithms currently implemented in commercial CGM devices, translating in suboptimal accuracy. The present review will analyze the issue of calibration algorithms for CGM, focusing exclusively on the commercially available glucose sensors.

Comprehensive pharmacokinetic model of insulin Glargine and other insulin formulations

In this paper, a comprehensive pharmacokinetic model for different insulin formulations including insulin Glargine is developed based on the model proposed by Trajanoski et al. (1993). Current models show limitations for insulin Glargine due to the appearance of an uncharacteristic peak in the concentration-time evolution of plasma insulin that does not coincide with real experimental data. This important limitation has been solved in this paper by introducing a new virtual insulin state called the bound state, in addition to the dimeric and hexameric ones. Trying to describe the retarded action of insulin Glargine, the modeling idea behind this approach is that immediately after the subcutaneous injection all the insulin resides in the bound state, and only then small amounts of insulin in the hexameric form disengage from the bound state. For the model evaluation different simulation results are compared. Using experimental data published by Lepore et al. (2000), the developed model turned out to be capable of at least qualitatively predicting the concentration-time profile of plasma insulin. Both exogenous insulin flow simulations and spatial diffusion simulations show the plausibility and correct implementation of the derived model. Considering all these simulation results, the here presented new pharmacokinetic model demonstrates to be able to reproduce real patient behavior simulating even complete insulin regimes including long-acting, intermediate and short-acting insulin formulations.

Real-Time Continuous Glucose Monitoring in an Intensive Care Unit: Better Accuracy in Patients with Septic Shock

OBJECTIVE This study assessed the accuracy of real-time continuous glucose monitoring system (RTCGMS) devices in an intensive care unit (ICU) to determine whether the septic status of the patient has any influence on the accuracy of the RTCGMS. SUBJECTS AND METHODS In total, 41 patients on insulin therapy were included. Patients were monitored for 72 h using RTCGMS. Arterial blood glucose (ABG) samples were obtained following the protocol established in the ICU. The results were evaluated using paired values (excluding those used for calibration) with the performance assessed using numerical accuracy. Nonparametric tests were used to determine statistically significant differences in accuracy. RESULTS In total, 956 ABG/RTCGMS pairs were analyzed. The overall median relative absolute difference (RAD) was 13.5%, and the International Organization for Standardization (ISO) criteria were 68.1%. The median RADs reported for patients with septic shock, with sepsis, and without sepsis were 11.2%, 14.3%, and 16.3%, respectively (P<0.05). Measurements meeting the ISO criteria were 74.5%, 65.6%, and 63.7% for patients with septic shock, with sepsis, and without sepsis, respectively (P<0.05). CONCLUSIONS The results showed that the septic status of patients influenced the accuracy of the RTCGMS in the ICU. Accuracy was significantly better in patients with septic shock in comparison with the other patient cohorts.

Robust Sliding Mode Closed-loop Glucose Control with Meal Compensation in Type 1 Diabetes Mellitus

Abstract This work addresses the design of a robust closed-loop plasma glucose controller for Type 1 Diabetes Mellitus patients. The feedback controller is based on Sliding Mode Control (SMC) while robust feedforward boluses to compensate food intake are calculated in a robust way by means of an interval glucose predictor that minimizes the risk of hypoglycaemia. The designed controller has been validated in a virtual environment following standard protocols. The resulting control algorithm shows a considerable robustness regarding intra-patient variability in insulin sensitivity as well as an enhanced ability to handle disturbance rejection. The International Diabetes Federation guidelines for glycaemia targets in Diabetes Mellitus are fulfilled by the designed control strategy.

Real-Time Glucose Estimation Algorithm for Continuous Glucose Monitoring Using Autoregressive Models

Background: Continuous glucose monitors (CGMs) present a problem of lack of accuracy, especially in the lower range, sometimes leading to missed or false hypoglycemia. A new algorithm is presented here aimed at improving the measurement accuracy and hypoglycemia detection. Its core is the estimation of blood glucose (BG) in real time (RT) from CGM intensity readings using autoregressive (AR) models. Methods: Eighteen patients with type 1 diabetes were monitored for three days (one at the hospital and two at home) using the CGMS® Gold. For these patients, BG samples were taken every 15 min for 2 h after meals and every half hour otherwise during the first day. The relationship between the current measured by the CGMS Gold and BG was learned by an AR model, allowing its RT estimation. New capillary glucose measurements were used to correct the model BG estimations. Results: A total of 563 paired points were obtained from BG and monitor readings to validate the new algorithm. 98.5% of paired points fell in zones A+B of the Clarke error grid analysis with the proposed algorithm. The overall mean and median relative absolute differences (RADs) were 9.6% and 6.7%. Measurements meeting International Organization for Standardization (ISO) criteria were 88.7%. In the hypoglycemic range, the mean and median RADs were 8.1% and 6.0%, and measurements meeting ISO criteria were 86.7%. The sensitivity and specificity with respect to hypoglycemia detection were 91.5% and 95.0%. Conclusions: The performance measured with both clinical and numerical accuracy metrics illustrates the improved accuracy of the proposed algorithm compared with values presented in the literature. A significant improvement in hypoglycemia detection was also observed.

Coordinated Basal-Bolus Infusion for Tighter Postprandial Glucose Control in Insulin Pump Therapy

Background: Basal and bolus insulin determination in intensive insulin therapy for type 1 diabetes mellitus (T1DM) are currently considered independently of each other. A new strategy that coordinates basal and bolus insulin infusion to cope with postprandial glycemia in pump therapy is proposed. Superior performance of this new strategy is demonstrated through a formal analysis of attainable performances in an in silico study. Methods: The set inversion via interval analysis algorithm has been applied to obtain the feasible set of basal and bolus doses that, for a given meal, mathematically guarantee a postprandial response fulfilling the International Diabetes Federation (IDF) guidelines (i.e., no hypoglycemia and 2 h postprandial glucose below 140 mg/dl). Hypoglycemia has been defined as a glucose value below 70 mg/dl. A 5 h time horizon has been considered for a 70 kg in silico T1DM subject consuming meals in the range of 30 to 80 g of carbohydrates. Results: The computed feasible sets demonstrate that current separated basal/bolus strategy dramatically limits the attainable performance. For a nominal basal of 0.8 IU/h leading to a basal glucose of approximately 100 mg/dl, IDF guidelines cannot be fulfilled for meals greater than 50 g of carbohydrates, independent of the bolus insulin computed. However, coordinating the basal and bolus insulin delivery can achieve this. A decrement of basal insulin during the postprandial period is required together with an increase in bolus insulin, in appropriate percentages, which is meal dependent. After 3 h, basal insulin can be restored to its nominal value. Conclusions: The new strategy meets IDF guidelines in a typical day, contrary to the standard basal/bolus strategy, yielding a mean 2 h postprandial glucose reduction of 36.4 mg/dl without late hypoglycemia. The application of interval analysis for the computation of feasible sets is demonstrated to be a powerful tool for the analysis of attainable performance in glucose control.

Controller Design Under Fuzzy Pole-Placement Specifications: An Interval Arithmetic Approach

This paper discusses fuzzy specifications for robust controller design, as a way to define different specification levels for different plants in a family and allow the control of performance degradation. Controller synthesis will be understood as mapping a fuzzy plant onto a desired fuzzy set of closed-loop specifications. In this context, a fuzzy plant is considered as a possibility distribution on a given plant space. In particular, pole placement in linear plants with fuzzy parametric uncertainty is discussed, although the basic idea is general and could be applied to other settings. In the case under consideration, the controller coefficients are the solution of a fuzzy linear system of equations with a particular semantics. Modal interval arithmetic is used to solve the system for each alpha-cut. The intersection of the solutions, if not empty, constitutes the solution to the robust control problem

A multiple local models approach to accuracy improvement in continuous glucose monitoring.

BACKGROUND Continuous glucose monitoring (CGM) devices estimate plasma glucose (PG) from measurements in compartments alternative to blood. The accuracy of currently available CGM is yet unsatisfactory and may depend on the implemented calibration algorithms, which do not compensate adequately for the differences of glucose dynamics between the compartments. Here we propose and validate an innovative calibration algorithm for the improvement of CGM performance. METHODS CGM data from GlucoDay(®) (A. Menarini, Florence, Italy) and paired reference PG have been obtained from eight subjects without diabetes during eu-, hypo-, and hyperglycemic hyperinsulinemic clamps. A calibration algorithm based on a dynamic global model (GM) of the relationship between PG and CGM in the interstitial space has been obtained. The GM is composed by independent local models (LMs) weighted and added. LMs are defined by a combination of inputs from the CGM and by a validity function, so that each LM represents to a variable extent a different metabolic condition and/or sensor-subject interaction. The inputs best suited for glucose estimation were the sensor current I and glucose estimations Ĝ, at different time instants [I(k), I(k)(-1), Ĝ(k)(-1)] (IIG). In addition to IIG, other inputs have been used to obtain the GM, achieving different configurations of the calibration algorithm. RESULTS Even in its simplest configuration considering only IIG, the new calibration algorithm improved the accuracy of the estimations compared with the manufacturers estimate: mean absolute relative difference (MARD)=10.8±1.5% versus 14.7±5.4%, respectively (P=0.012, by analysis of variance). When additional exogenous signals were considered, the MARD improved further (7.8±2.6%, P<0.05). CONCLUSIONS The LM technique allows for the identification of intercompartmental glucose dynamics. Inclusion of these dynamics into the calibration algorithm improves the accuracy of PG estimations.

Safety Auxiliary Feedback Element for the Artificial Pancreas in Type 1 Diabetes

The artificial pancreas aims at the automatic delivery of insulin for glycemic control in patients with type 1 diabetes, i.e., closed-loop glucose control. One of the challenges of the artificial pancreas is to avoid controller overreaction leading to hypoglycemia, especially in the late postprandial period. In this study, an original proposal based on sliding mode reference conditioning ideas is presented as a way to reduce hypoglycemia events induced by a closed-loop glucose controller. The method is inspired in the intuitive advantages of two-step constrained control algorithms. It acts on the glucose reference sent to the main controller shaping it so as to avoid violating given constraints on the insulin-on-board. Some distinctive features of the proposed strategy are that 1) it provides a safety layer which can be adjusted according to medical criteria; 2) it can be added to closed-loop controllers of any nature; 3) it is robust against sensor failures and overestimated prandial insulin doses; and 4) it can handle nonlinear models. The method is evaluated in silico with the ten adult patients available in the FDA-accepted UVA simulator.

Using Support Vector Machines to Detect Therapeutically Incorrect Measurements by the MiniMed CGMS

Background: Current continuous glucose monitors have limited accuracy mainly in the low range of glucose measurements. This lack of accuracy is a limiting factor in their clinical use and in the development of the so-called artificial pancreas. The ability to detect incorrect readings provided by continuous glucose monitors from raw data and other information supplied by the monitor itself is of utmost clinical importance. In this study, support vector machines (SVMs), a powerful statistical learning technique, were used to detect therapeutically incorrect measurements made by the Medtronic MiniMed CGMS®. Methods: Twenty patients were monitored for three days (first day at the hospital and two days at home) using the MiniMed CGMS. After the third day, the monitor data were downloaded to the physicians computer. During the first 12 hours, the patients stayed in the hospital, and blood samples were taken every 15 minutes for two hours after meals and every 30 minutes otherwise. Plasma glucose measurements were interpolated using a cubic method for time synchronization with simultaneous MiniMed CGMS measurements every five minutes, obtaining a total of 2281 samples. A Gaussian SVM classifier trained on the monitors electrical signal and glucose estimation was tuned and validated using multiple runs of k-fold cross-validation. The classes considered were Clarke error grid zones A+B and C+D+E. Results: After ten runs of ten-fold cross-validation, an average specificity and sensitivity of 92.74% and 75.49%, respectively, were obtained (see Figure 4). The average correct rate was 91.67%. Conclusions: Overall, the SVM performed well, in spite of the somewhat low sensitivity. The classifier was able to detect the time intervals when the monitors glucose profile could not be trusted due to incorrect measurements. As a result, hypoglycemic episodes missed by the monitor were detected.