Accuracy evaluation of a new real-time continuous glucose monitoring algorithm in hypoglycemia

A Critical Discussion of Alert Evaluations in the Context of Continuous Glucose Monitoring System Performance

Many continuous glucose monitoring (CGM) systems provide functionality which alerts users of potentially unwanted glycemic conditions. These alerts can include glucose threshold alerts to call the user's attention to hypoglycemia or hyperglycemia, predictive alerts warning about impeding hypoglycemia or hyperglycemia, and rate-of-change alerts. A recent review identified 129 articles about CGM performance studies, of which approximately 25% contained alert evaluations. In some studies, real alerts were assessed; however, most of these studies retrospectively determined the timing of CGM alerts because not all CGM systems record alerts which necessitates manual documentation. In contrast to assessment of real alerts, retrospective determination allows assessment of a variety of alert settings for all three types of glycemic condition alerts. Based on the literature and the Clinical and Laboratory Standards Institute's POCT05 guideline, two common approaches to threshold alert evaluation were identified, one value-based and one episode-based approach. In this review, a critical discussion of the two approaches, including a post hoc analysis of clinical study data, indicates that the episode-based approach should be preferred over the value-based approach. For predictive alerts, fewer results were found in the literature, and retrospective determination of CGM alert timing is complicated by the prediction algorithms being proprietary information. Rate-of-change alert evaluations were not reported in the identified literature, and POCT05 does not contain recommendations for assessment. A possible approach is discussed including post hoc analysis of clinical study data. To conclude, CGM systems should record alerts, and the episode-based approach to alert evaluation should be preferred.

Clinical Performance Evaluation of Continuous Glucose Monitoring Systems: A Scoping Review and Recommendations for Reporting

The use of different approaches for design and results presentation of studies for the clinical performance evaluation of continuous glucose monitoring (CGM) systems has long been recognized as a major challenge in comparing their results. However, a comprehensive characterization of the variability in study designs is currently unavailable. This article presents a scoping review of clinical CGM performance evaluations published between 2002 and 2022. Specifically, this review quantifies the prevalence of numerous options associated with various aspects of study design, including subject population, comparator (reference) method selection, testing procedures, and statistical accuracy evaluation. We found that there is a large variability in nearly all of those aspects and, in particular, in the characteristics of the comparator measurements. Furthermore, these characteristics as well as other crucial aspects of study design are often not reported in sufficient detail to allow an informed interpretation of study results. We therefore provide recommendations for reporting the general study design, CGM system use, comparator measurement approach, testing procedures, and data analysis/statistical performance evaluation. Additionally, this review aims to serve as a foundation for the development of a standardized CGM performance evaluation procedure, thereby supporting the goals and objectives of the Working Group on CGM established by the Scientific Division of the International Federation of Clinical Chemistry and Laboratory Medicine.

Minimally invasive electrochemical continuous glucose monitoring sensors: Recent progress and perspective

Diabetes and its complications are seriously threatening the health and well-being of hundreds of millions of people. Glucose levels are essential indicators of the health conditions of diabetics. Over the past decade, concerted efforts in various fields have led to significant advances in glucose monitoring technology. In particular, the rapid development of continuous glucose monitoring (CGM) based on electrochemical sensing principles has great potential to overcome the limitations of self-monitoring blood glucose (SMBG) in continuously tracking glucose trends, evaluating diabetes treatment options, and improving the quality of life of diabetics. However, the applications of minimally invasive electrochemical CGM sensors are still limited owing to the following aspects: i) invasiveness, ii) short lifespan, iii) biocompatibility, and iv) calibration and prediction. In recent years, the performance of minimally invasive electrochemical CGM systems (CGMSs) has been significantly improved owing to breakthrough developments in new materials and key technologies. In this review, we summarize the history of commercial CGMSs, the development of sensing principles, and the research progress of minimally invasive electrochemical CGM sensors in reducing the invasiveness of implanted probes, maintaining enzyme activity, and improving the biocompatibility of the sensor interface. In addition, this review also introduces calibration algorithms and prediction algorithms applied to CGMSs and describes the application of machine learning algorithms for glucose prediction.

Hypoglycaemia detection and prediction techniques: A systematic review on the latest developments

The main objective of diabetes control is to correct hyperglycaemia while avoiding hypoglycaemia, especially in insulin-treated patients. Fear of hypoglycaemia is a hurdle to effective correction of hyperglycaemia because it promotes under-dosing of insulin. Strategies to minimise hypoglycaemia include education and training for improved hypoglycaemia awareness and the development of technologies to allow their early detection and thus minimise their occurrence. Patients with impaired hypoglycaemia awareness would benefit the most from these technologies. The purpose of this systematic review is to review currently available or in-development technologies that support detection of hypoglycaemia or hypoglycaemia risk, and identify gaps in the research. Nanomaterial use in sensors is a promising strategy to increase the accuracy of continuous glucose monitoring devices for low glucose values. Hypoglycaemia is associated with changes on vital signs, so electrocardiogram and encephalogram could also be used to detect hypoglycaemia. Accuracy improvements through multivariable measures can make already marketed galvanic skin response devices a good noninvasive alternative. Breath volatile organic compounds can be detected by dogs and devices and alert patients at hypoglycaemia onset, while near-infrared spectroscopy can also be used as a hypoglycaemia alarms. Finally, one of the main directions of research are deep learning algorithms to analyse continuous glucose monitoring data and provide earlier and more accurate prediction of hypoglycaemia. Current developments for early identification of hypoglycaemia risk combine improvements of available 'needle-type' enzymatic glucose sensors and noninvasive alternatives. Patient usability will be essential to demonstrate to allow their implementation for daily use in diabetes management.

Feature-Based Machine Learning Model for Real-Time Hypoglycemia Prediction

BACKGROUND

Hypoglycemia is a serious health concern in youth with type 1 diabetes (T1D). Real-time data from continuous glucose monitoring (CGM) can be used to predict hypoglycemic risk, allowing patients to take timely intervention measures.

METHODS

A machine learning model is developed for probabilistic prediction of hypoglycemia (<70 mg/dL) in 30- and 60-minute time horizons based on CGM datasets obtained from 112 patients over a range of 90 days consisting of over 1.6 million CGM values under normal living conditions. A comprehensive set of features relevant for hypoglycemia are developed and a parsimonious subset with most influence on predicting hypoglycemic risk is identified. Model performance is evaluated both with and without contextual information on insulin and carbohydrate intake.

RESULTS

The model predicted hypoglycemia with >91% sensitivity for 30- and 60-minute prediction horizons while maintaining specificity >90%. Inclusion of insulin and carbohydrate data yielded performance improvement for 60-minute but not for 30-minute predictions. Model performance was highest for nocturnal hypoglycemia (~95% sensitivity). Shortterm (less than one hour) and medium-term (one to four hours) features for good prediction performance are identified.

CONCLUSIONS

Innovative feature identification facilitated high performance for hypoglycemia risk prediction in pediatric youth with T1D. Timely alerts of impending hypoglycemia may enable proactive measures to avoid severe hypoglycemia and achieve optimal glycemic control. The model will be deployed on a patient-facing smartphone application in an upcoming pilot study.

Benefits and Limitations of MARD as a Performance Parameter for Continuous Glucose Monitoring in the Interstitial Space

High-quality performance of medical devices for glucose monitoring is important for a safe and efficient usage of this diagnostic option by patients with diabetes. The mean absolute relative difference (MARD) parameter is used most often to characterize the measurement performance of systems for continuous glucose monitoring (CGM). Calculation of this parameter is relatively easy and comparison of the MARD numbers between different CGM systems appears to be straightforward on the first glance. However, a closer look reveals that a number of complex aspects make interpretation of the MARD numbers provided by the manufacturer for their CGM systems difficult. In this review, these aspects are discussed and considerations are made for a systematic and appropriate evaluation of the MARD in clinical trials. The MARD should not be used as the sole parameter to characterize CGM systems, especially when it comes to nonadjunctive usage of such systems.

Design of a Real-time Self-adjusting Calibration Algorithm to Improve the Accuracy of Continuous Blood Glucose Monitoring

The aim of this study is to establish a real-time self-adjusting calibration algorithm to compensate for signal drift and sensitivity attenuation of subcutaneous implantable glucose sensors. A real-time self-adjusting in vivo calibration method was designed based on the one-point calibration model. The current signal was compensated in real-time and the sensitivity was calibrated regularly. The least squares method was used to fit the initial parameters of the model, and then, the in vivo monitored current data was calibrated. Comparing with the mean absolute relative difference (MARD) of the blood glucose concentration by the traditional one-point calibration model (22.85 ± 5.76%), the MARD of the blood glucose concentration calibrated by the real-time self-adjusting in vivo calibration method was 6.28 ± 2.31%. The accuracy of the dynamic blood glucose monitoring was effectively improved. This calibration algorithm could compensate the signal drift in real time and correct sensitivity regularly to improve the accuracy of dynamic glucose monitoring, thus significantly enhancing diabetic self-management.

Comparison of Continuous Glucose Monitoring Accuracy Between Abdominal and Upper Arm Insertion Sites

Background: The aim was to compare the accuracy of the Dexcom^® G4 Platinum continuous glucose monitor (CGM) sensor inserted on the upper arm and the abdomen in adults. Methods: Fourteen adults with type 1 diabetes wore two CGMs, one placed on the upper arm and one placed on the abdomen. Three in-clinic visits of 5 h with YSI (2300 STAT, Yellow Springs Instrument) measurements as comparator were performed. Each visit was followed by 4 days with seven-point self-monitoring of blood glucose (SMBG) in free-living conditions. Accuracy analyses on the paired CGM-YSI and CGM-SMBG measurements of the two CGM sensors were performed. Results: Using YSI as comparator, the overall Mean Absolute Relative Difference (MARD) for the CGM_abd was 12.3% and CGM_arm was 12.0%. The percentage of the CGM measurements in zone A of Clarke error grid analysis for the CGM_abd was 85.6% and CGM_arm was 86.0%. The hypoglycemia sensitivity for the CGM_abd and CGM_arm was 69.3%. Using SMBG as comparator, the overall MARD for the CGM_abd was 12.5% and CGM_arm was 12.0%. The percentage of the CGM measurements in zone A for the CGM_abd was 84.1% and the CGM_arm was 85.0%. The hypoglycemia sensitivity for the CGM_abd was 60.0% and the CGM_arm was 71.1%. All the P-values from the comparisons between the accuracy of CGM_abd and CGM_arm were >0.05. Conclusion: The accuracy of a Dexcom G4 Platinum CGM sensor placed on the upper arm was not different from the accuracy of the sensor placed on the abdomen in adults with type 1 diabetes.

An ARIMA Model With Adaptive Orders for Predicting Blood Glucose Concentrations and Hypoglycemia

The continuous glucose monitoring system is an effective tool, which enables the users to monitor their blood glucose (BG) levels. Based on the continuous glucose monitoring (CGM) data, we aim at predicting future BG levels so that appropriate actions can be taken in advance to prevent hyperglycemia or hypoglycemia. Due to the time-varying nonstationarity of CGM data, verified by Augmented Dickey-Fuller test and analysis of variance, an autoregressive integrated moving average (ARIMA) model with an adaptive identification algorithm of model orders is proposed in the prediction framework. Such identification algorithm adaptively determines the model orders and simultaneously estimates the corresponding parameters using Akaike Information Criterion and least square estimation. A case study is conducted with the CGM data of diabetics under daily living conditions to analyze the prediction performance of the proposed model together with the early hypoglycemic alarms. Results show that the proposed model outperforms the adaptive univariate model and ARIMA model.

A Single-Use, Self-Powered, Paper-Based Sensor Patch for Detection of Exercise-Induced Hypoglycemia

We report a paper-based self-powered sensor patch for prevention and management of exercise-induced hypoglycemia. The article describes the fabrication, in vitro, and in vivo characterization of the sensor for glucose monitoring in human sweat. This wearable, non-invasive, single-use biosensor integrates a vertically stacked, paper-based glucose/oxygen enzymatic fuel cell into a standard Band-Aid adhesive patch. The paper-based device attaches directly to skin, wicks sweat by using capillary forces to a reservoir where chemical energy is converted to electrical energy, and monitors glucose without external power and sophisticated readout instruments. The device utilizes (1) a 3-D paper-based fuel cell configuration, (2) an electrically conducting microfluidic reservoir for a high anode surface area and efficient mass transfer, and (3) a direct electron transfer between glucose oxidase and anodes for enhanced electron discharge properties. The developed sensor shows a high linearity of current at 0.02–1.0 mg/mL glucose centration (R² = 0.989) with a high sensitivity of 1.35 µA/mM.

Toward Big Data Analytics: Review of Predictive Models in Management of Diabetes and Its Complications

Diabetes is one of the top priorities in medical science and health care management, and an abundance of data and information is available on these patients. Whether data stem from statistical models or complex pattern recognition models, they may be fused into predictive models that combine patient information and prognostic outcome results. Such knowledge could be used in clinical decision support, disease surveillance, and public health management to improve patient care. Our aim was to review the literature and give an introduction to predictive models in screening for and the management of prevalent short- and long-term complications in diabetes. Predictive models have been developed for management of diabetes and its complications, and the number of publications on such models has been growing over the past decade. Often multiple logistic or a similar linear regression is used for prediction model development, possibly owing to its transparent functionality. Ultimately, for prediction models to prove useful, they must demonstrate impact, namely, their use must generate better patient outcomes. Although extensive effort has been put in to building these predictive models, there is a remarkable scarcity of impact studies.

Effect of Continuous Glucose Monitoring Accuracy on Clinicians' Retrospective Decision Making in Diabetes: A Pilot Study

BACKGROUND

The use of continuous glucose monitoring (CGM) in clinical decision making in diabetes could be limited by the inaccuracy of CGM data when compared to plasma glucose measurements. The aim of the present study is to investigate the impact of CGM numerical accuracy on the precision of diabetes treatment adjustments.

METHOD

CGM profiles with maximum 5-day duration from 12 patients with type 1 diabetes treated with a basal-bolus insulin regimen were processed by 2 CGM algorithms, with the accuracy of algorithm 2 being higher than the accuracy of algorithm 1, using the median absolute relative difference (MARD) as the measure of accuracy. During 2 separate and similar occasions over a 1-month interval, 3 clinicians reviewed the processed CGM profiles, and adjusted the dose level of basal and prandial insulin. The precision of the dosage adjustments were defined in terms of the interclinician agreement and the intraclinician reproducibility of the decisions. The Cohen's kappa coefficient was used to assess the precision of the decisions. The study was based on retrospective and blind CGM data.

RESULTS

For the interclinician agreement, in the first occasion, the kappa of algorithm 1 was .32, and that of algorithm 2 was .36. For the interclinician agreement, in the second occasion, the kappas of algorithms 1 and 2 were .17 and .22, respectively. For the intraclinician reproducibility of the decisions, the kappas of algorithm 1 were .35, .22, and .80 and the kappas of algorithm 2 were .44, .52, and .32, for the 3 clinicians, respectively. For the interclinician agreement, the relative kappa change from algorithm 1 to algorithm 2 was 86.06%, and for the intraclinician reproducibility, the relative kappa change from algorithm 1 to algorithm 2 was 53.99%.

CONCLUSIONS

Results indicated that the accuracy of CGM algorithms might potentially affect the precision of the CGM-based insulin adjustments for type 1 diabetes patients. However, a larger study with several clinical centers, with higher number of clinicians and patients is required to validate the impact of CGM accuracy on decisions precision.