Continuous Glucose Monitoring Signals Research Articles

Abstract P600: Beyond Mean Glucose: Development of Novel, Reproducible Continuous Glucose Monitoring Metrics

Introduction: Continuous glucose monitoring (CGM) devices record blood glucose every few minutes over a maximum 2-week wear period. Despite the richness of these data, existing first and second moment statistics (such as mean glucose) fail to capture the “shape” of variation in the CGM time-series. Objective: We developed novel, reproducible CGM shape metrics using functionals of the periodogram, a decomposition of CGM signal ( Figure 1A ) variance, to comprehensively characterize glucose patterns. Methods: We analyzed CGM data from N = 160 HYPNOS clinical trial participants with type 2 diabetes who wore CGM sensors twice, 3 months apart. We first calculated log-transformed periodograms for each person-period. To these decompositions, we fit piece-wise linear models over 3 frequency ranges: less than 1/24 hrs -1 (long term patterns), 1/24 to 2/5 hrs -1 (daily diet patterns), and greater than 2/5 hrs -1 (immediate fluctuations due to food). Slopes of fit segments, value at first frequency, and values at segment midpoints were extracted as our initial metrics ( Figure 1B ). We estimated and compared the within-person test-retest correlation for our shape measures against existing CGM and glucose metrics: mean glucose, time-in-range, and HhbA1c. Results: The highest raw correlation among the new shape metrics (r = 0.737) was comparable to that of the existing metrics (r = 0.798) ( Figure 1C ). Even after adjusting the shape metrics for existing metrics, there was a maximum test-retest correlation of r = 0.778. The shape metrics also carried information distinct from existing metrics. According to mixed effect models fit to data from both wear periods, the shape metrics explain less than 16% of variance in any existing metric when controlled for the other existing metrics. Conclusion: These new shape metrics carry reproducible information distinct from that of standard CGM metrics and HbA1c. The next step is to evaluate whether these novel shape metrics, which leverage the granularity of CGM data, can be linked to clinical endpoints.

ODP210 Impact of Continuous Glucose Monitoring (CGM) in the Intensive Care Units (ICUs): Accuracy, Reliability, Feasibility and Nurse Acceptance

Abstract Background The COVID-19 pandemic led to crisis-level shortage of hospital beds and staff, predominantly in the ICUs. Minimization of patient contact moved the FDA to issue a statement that allowed CGMs in hospitals, which were quickly adopted by ICUs. We conducted a study to better understand the use of CGM in ICUs and test for accuracy, reliability feasibility, and acceptance by care providers. Methods Patients were eligible for CGM if they had diabetes mellitus and concomitant COVID-19 infection that required ICU level of care. CGM signals were received by smart phones that were placed in a transparent biohazard bag outside patients’ rooms. Nurses were educated on the interpretation of glucose trends, alerts, and calibration. CGM alerts were set for glucose readings below 100mg/dL and above or equal to 250mg/dL. Nursing staff could obtain point-of-care (POC) glucose as scheduled and for CGM alarms. Mean Absolute Relative Difference (MARD) values were calculated for accuracy. Nurses were invited to respond to a questionnaire with 5-point Likert scales to assess their perception of CGM use, and the potential decrease in the amount of personal protective equipment (PPE) needed in the care of patients on CGM. A question on the estimated time it takes to donn and doff PPE was also included. Results 26 patients were included in the study; average age 60.8 + 13.5 years, 7 were female and 12 belonged to underrepresented minorities (Hispanic, non-Hispanic Black, or Asian); mean HbA1c was 9.2 + 2.6%. 7 patients were on intravenous insulin infusion, 6 were intubated and 5 received vasopressors. Average duration on CGM was 6 + 3.75 days.168 paired serum glucose-CGM measurements and 259 paired POC-CGM measurements were used. MARD between serum glucose and CGM measurements was 11.2%; MARD between POC glucose and CGM measurement was 11.6%. Questionnaires from 33 nursing staff were analyzed; the most frequent answer (mode) for time to donn and doff PPE was 10 minutes. The vast majority of answers were "agree" (21.2%) or "strongly agree" (78.8%) to the statements that CGM was easy to use and interpret and during the pandemic staff felt protected with use of CGM. Likewise, the answers were "agree" (18.8%) or "strongly agree" (81.8%) that CGMs could replace POC glucose in COVID-19 patients and that there was a lower need for PPE. Conclusion There is not an established acceptable MARD for patients with COVID-19 in an ICU setting. Wollersheim et al. suggested that CGM with a MARD of <14% is acceptable for ICU patients. The MARD we found in our study was well within this range. We concluded that use of CGM in the ICU setting has acceptable accuracy and is reliable, feasible and well accepted by the nursing staff. Presentation: No date and time listed

LBODP061 A Pilot Trial Of Continuous Glucose Monitoring Technology In The Perioperative Period

Abstract Objective To evaluate use of continuous glucose monitoring (CGM) technology in the perioperative setting by comparing CGM with concurrent point of care fingerstick blood glucose (BG) or arterial BG values, as indicated, in patients with hyperglycemia and/or diabetes mellitus undergoing surgery. To assess if grounding device correlated with loss of CGM signal as a prior study of CGM in the perioperative period noted intermittent signal loss during operative time. Methods A prospective cohort analysis of 94 participants with hyperglycemia and/or diabetes mellitus undergoing surgery of 3 hours duration was conducted using Dexcom G6 and/or Freestyle Libre 2. 0 CGM. CGMs were placed preoperatively and compared to standard of care BG checks in the perioperative setting. Correction insulin was administered based on fingerstick glucoses. Of those consented, 18 were excluded resulting in 76 enrolled subjects. Results Preliminary data for use of CGM in perioperative period analyzed 50 participants with Freestyle Libre 2. 0 CGM, 20 participants with Dexcom G6 CGM, and 6 participants with both devices worn simultaneously. Lost sensor data occurred for 3 participants wearing Dexcom G6 CGM and 10 participants wearing Freestyle Libre 2. 0 CGM, which affected 2 of the participants wearing both devices simultaneously. The absolute difference of Dexcom G6 CGM glucose readings compared to BG readings was 29.17 mg/dL (SD = 22.15). The absolute difference of Freestyle Libre 2. 0 CGM glucose readings compared to BG readings was 26.94 mg/dL (SD = 24.16). There was not a significant difference in averages of absolute change in glucose when comparing Dexcom G6 CGM to Freestyle Libre 2. 0 CGM when utilized simultaneously. Grounding devices utilized Megadyne grounding pad for use of bipolar cautery, monopolar cautery, and surgeries with da Vinci robot. Of the 13 participants with lost data, 3 patients had bipolar cautery reported as being used at least once and 10 participants with lost data utilized monopolar cautery. None of the 6 participants with surgery utilizing da Vinci robot had loss of CGM data. Conclusion Both Dexcom G6 and Freestyle Libre 2. 0 CGMs were able to be utilized in the perioperative setting and generally functioned well. Use of CGM provided more glycemic data and better characterized glycemic trends than individual BG readings. It is anticipated that this technology could improve prevention of hypoglycemic events and better glycemic control in the perioperative setting. Use of CGMs in the perioperative setting is a novel area of application that warrants further evaluation of this technology. Presentation: No date and time listed

Toward an Optimal Definition of Hypoglycemia with Continuous Glucose Monitoring

Background and ObjectiveAs continuous glucose monitoring (CGM) becomes common in research and clinical practice, there is a need to understand how CGM-based hypoglycemia relates to hypoglycemia episodes defined conventionally as patient reported hypoglycemia (PRH). Data show that CGM identify many episodes of low interstitial glucose (LIG) that are not experienced by patients, and so the aim of this study is to use different PRH simulations to optimize CGM parameters of threshold (h) and duration (d) to provide the best PRH detection performance. MethodsThe algorithm uses particle Markov chain Monte Carlo optimization to identify the optimal h and d which maximize an objective function for detecting PRH. We tested our algorithm by creating three different cases of PRH simulations. ResultsWe added three types of simulated PRH events to 10 weeks of anonymized CGM data from 96 type 1 diabetes people to see if the algorithm can detect the optimal parameters set out in the simulations. In simulation 1, we changed the locations of PRHs with respect to LIG episodes in the CGM signal to simulate random optimal LIG parameters for every individual. In simulation 2, the PRHs are CGM glucose <3.9 mmol/L followed by at least 20 min of rise > 0.11 mmol/L/min. Simulation 3 is like simulation 2 but with glucose threshold of 3.0 mmol/L. The median [interquartile range] of deviation between the optimized (found by the algorithm) and the optimal (known) h and d are −0.07% [−0.4, 1.9] and −1.3% [−5.9, 6.8], respectively across the subjects for simulation 1. The mean [min max] of the optimized LIG parameters are h = 3.8 [3.7, 3.8] mmol/L and d = 12 [10, 14] min for simulation 2 and they are h = 3.0 [2.9, 3] mmol/L and d = 10 [8, 14] min for simulation 3 across a 10-fold cross validation. ConclusionsThis work demonstrates the feasibility of the algorithm to find the best-fit definition of CGM-based hypoglycemia for PRH detection. In a prospective clinical study collecting CGM and PRH, the current algorithm will be used to optimize the definition of hypoglycemia with respect to PRH with the ambition of using the resulted definition as a surrogate for PRH in clinical practice.

Multilevel clustering approach driven by continuous glucose monitoring data for further classification of type 2 diabetes

IntroductionMining knowledge from continuous glucose monitoring (CGM) data to classify highly heterogeneous patients with type 2 diabetes according to their characteristics remains unaddressed. A refined clustering method that retrieves hidden...

975-P: Effect of Late Bolus Injections on Glycemic Variability Studied by Connected Pens

Mealtime injection of bolus insulin is critical for maintaining good glycemic control in patients with type 1 diabetes (T1D). It has previously been shown in a real-world setting that patients using connected pens had fewer missed mealtime injections of bolus doses as well as improved time-in-range (TIR) compared to baseline. This subgroup post-hoc study examined the effect of using connected pens on timing of bolus doses in adults and the relationship between timing of the insulin dose and glycemic variability. In a non-interventional study, conducted at 12 Swedish diabetes clinics, connected pens with bolus and/or basal insulin were prescribed to adults with T1D. Injection- and continuous glucose monitoring (CGM) data were collected at routine clinic visits. CGM data were also uploaded at home. These data were used by healthcare providers (HCPs) and patients during visits at the clinic. Meals were subsequently detected based on the CGM signals using the clinically validated Glucose Rate Increase Detector algorithm. A bolus dose given 60-120 minutes after the estimated start of the meal was defined as a late-bolus dose. In total 96 patients with 2238 days of acceptable CGM coverage (≥ 70%) were included in the analysis. The mean number of late-bolus doses decreased from 0.32 per day (95% CI [0.26;0.39]) at baseline to 0.18 per day [0.13;0.27] at follow-up, corresponding to a 42% reduction (p = 0.005). A significant correlation between the timing of the bolus dose and the coefficient of variation (CV) of the CGM signal was found; each 10-minute delay of the bolus dose was associated with an increase of 0.5% CV (p &amp;lt; 0.0001) on average. This study confirms that timely injection of bolus doses relative to meals is correlated with a more stable glycemic control. Also, the results indicate that timely bolus dosing can be a challenge in everyday life but that data from connected pens may help support patient behavior through improved interaction with HCP. Disclosure J.H. Jendle: Advisory Panel; Self; Abbott, AstraZeneca, Boehringer Ingelheim International GmbH, Eli Lilly and Company, Medtronic, Novo Nordisk A/S, Sanofi. Research Support; Self; Novo Nordisk A/S. Speaker’s Bureau; Self; AstraZeneca, Boehringer Ingelheim International GmbH, Eli Lilly and Company, Novo Nordisk A/S, Sanofi. N.V. Hartvig: Employee; Self; Novo Nordisk A/S. A. Kaas: Employee; Self; Novo Nordisk A/S. Stock/Shareholder; Self; Novo Nordisk A/S. J. Moller: Employee; Self; Novo Nordisk A/S. Stock/Shareholder; Self; Novo Nordisk A/S. A.M. Mårdby: Employee; Self; Novo Nordisk A/S. Stock/Shareholder; Self; Novo Nordisk A/S. S. Catrina: None.

Long-Term Glucose Forecasting Using a Physiological Model and Deconvolution of the Continuous Glucose Monitoring Signal.

(1) Objective: Blood glucose forecasting in type 1 diabetes (T1D) management is a maturing field with numerous algorithms being published and a few of them having reached the commercialisation stage. However, accurate long-term glucose predictions (e.g., >60 min), which are usually needed in applications such as precision insulin dosing (e.g., an artificial pancreas), still remain a challenge. In this paper, we present a novel glucose forecasting algorithm that is well-suited for long-term prediction horizons. The proposed algorithm is currently being used as the core component of a modular safety system for an insulin dose recommender developed within the EU-funded PEPPER (Patient Empowerment through Predictive PERsonalised decision support) project. (2) Methods: The proposed blood glucose forecasting algorithm is based on a compartmental composite model of glucose–insulin dynamics, which uses a deconvolution technique applied to the continuous glucose monitoring (CGM) signal for state estimation. In addition to commonly employed inputs by glucose forecasting methods (i.e., CGM data, insulin, carbohydrates), the proposed algorithm allows the optional input of meal absorption information to enhance prediction accuracy. Clinical data corresponding to 10 adult subjects with T1D were used for evaluation purposes. In addition, in silico data obtained with a modified version of the UVa-Padova simulator was used to further evaluate the impact of accounting for meal absorption information on prediction accuracy. Finally, a comparison with two well-established glucose forecasting algorithms, the autoregressive exogenous (ARX) model and the latent variable-based statistical (LVX) model, was carried out. (3) Results: For prediction horizons beyond 60 min, the performance of the proposed physiological model-based (PM) algorithm is superior to that of the LVX and ARX algorithms. When comparing the performance of PM against the secondly ranked method (ARX) on a 120 min prediction horizon, the percentage improvement on prediction accuracy measured with the root mean square error, A-region of error grid analysis (EGA), and hypoglycaemia prediction calculated by the Matthews correlation coefficient, was , , and , respectively. Although showing a trend towards improvement, the addition of meal absorption information did not provide clinically significant improvements. (4) Conclusion: The proposed glucose forecasting algorithm is potentially well-suited for T1D management applications which require long-term glucose predictions.

Analysis of "Simple Post-Processing of Continuous Glucose Monitoring Measurements Improves Endpoints in Clinical Trials".

Jensen et al used continuous glucose monitoring (CGM) data of the Dexcom G4 Platinum (DG4P) sensor obtained in a clinical efficacy and safety study of Novo Nordisk's new insulin Fiasp® to calculate the CGM time delay versus plasma glucose (PG) and self-measured blood glucose (SMBG) measurements (9-10 min). Shifting the CGM signal by 9 min backward in time versus PG and SMBG data improved the analytical accuracy of the DG4P sensor and the reliability of clinical research endpoint (hypoglycemia, postprandial glucose increments) detection. Since this method takes advantage of post-processing of CGM data, it is particularly suited for the optimization of data processing in clinical studies. In contrast, real-time corrections of time delays need predictive algorithms.

126-LB: Improved Insulin Adherence after Introduction of a Smart Connected Insulin Pen

An association between missed insulin injections and its impact on HbA1c levels has been established. The smart connected NovoPen® 6 captures and allows visualization of insulin injections (date and time of injection and number of units) and thus has the potential to move dialog between patients and healthcare professionals (HCPs) away from guessing about doses taken, missed doses and optimal injection time in relation to meals. This non-interventional study investigated whether the use of NovoPen® 6 influenced the behavior of patients with type 1 diabetes (T1D) in terms of missed bolus dose (MBD) injections. Patients were recruited from 12 Swedish diabetes clinics. At baseline they received a NovoPen® 6 for bolus insulin injections. At each HCP visit, pen data were downloaded at the clinic. Follow-up was after ≥5 HCP visits. Adults with T1D (n=81) using continuous glucose monitoring (CGM) and NovoPen® 6 for bolus injections were included in the analyses. The frequency of MBD was analyzed using the GRID algorithm to detect meals from the CGM signal. MBD was defined as meals with no bolus injection within -15 to +60 minutes from the start of the meal, as detected by the algorithm. The change in number of MBD from baseline to ≥5 HCP visits was analyzed using a mixed model. A significant decrease of 43% in the average daily number of MBD was observed from baseline to after ≥5 HCP visits (median time in study 6 months) from 0.74 (95% CI [0.62; 0.88]) to 0.42 (95% CI [0.30; 0.60]) (p=0.002). Based on the assumption that patients have three main meals per day, this corresponded to a decrease from 25% to 14% in MBD. The results indicated that patients achieved more well-dosed meals, even though these are not detected by the GRID algorithm, because of a lower CGM response. These real-world findings confirm that MBD is a problem for patients with T1D and that a smart connected pen can support good injection behavior, with fewer missed and more well-dosed mealtime injections. This could subsequently lead to better glycemic control. Disclosure P. Adolfsson: Advisory Panel; Self; Eli Lilly and Company, Novo Nordisk A/S, Roche Pharma. Employee; Self; Region Halland. Research Support; Self; Region Halland. N.V. Hartvig: Employee; Self; Novo Nordisk A/S. Stock/Shareholder; Self; Novo Nordisk A/S. A. Kaas: Employee; Self; Novo Nordisk A/S. Stock/Shareholder; Self; Novo Nordisk A/S. N. Nygård Knudsen: Employee; Self; Novo Nordisk A/S. A. Mårdby: Employee; Self; Novo Nordisk A/S. Stock/Shareholder; Self; Novo Nordisk A/S. J.B. Møller: Employee; Self; Novo Nordisk A/S. Stock/Shareholder; Self; Novo Nordisk A/S. J. Hellman: Advisory Panel; Self; Abbott, Boehringer Ingelheim Pharmaceuticals, Inc., Eli Lilly and Company, Merck Sharp &amp; Dohme Corp., Novo Nordisk A/S, Sanofi. Consultant; Self; Boehringer Ingelheim Pharmaceuticals, Inc., Novo Nordisk A/S, Sanofi. Funding Novo Nordisk

A concurrent fault and meal detection method based on dynamics analysis for continuous glucose monitoring sensor

The quality of continuous glucose monitoring (CGM) signals influences the efficiency of the diabetes treatment. Faults of the CGM sensors may lead to inaccurate treatment strategy and even threaten the lives of the diabetic patients. The transient loss of sensitivity of sensor is a common failure which has a small magnitude including early changing and the slow developing. It is difficult to be detected since it is sometimes buried by nonstationary trends caused by the unannounced meals. In order to detect transient loss of sensitivity and unannounced meals accurately, a concurrent detection method is proposed based on canonical variate analysis (CVA) and slow feature analysis (SFA) for analysis of dynamic information, including the time serial correlation and changing speed of CGM signals. First, the CVA is used to analyze the time series correlation of CGM signals which may be broken by either sensor faults or unannounced meals. Further, to distinguish faults and unannounced meals, the SFA is proposed to extract temporal features since faults and meals will cause different changing speeds of CGM signals. In this way, a concurrent detection method is established to online detect and distinguish the real faults and unannounced meals. The effectiveness of the proposed method is verified using data collected from thirty in silico subjects with type 1 diabetes.

Knowledge-driven dictionaries for sparse representation of continuous glucose monitoring signals.

Continuous glucose monitoring (CGM) of patients with diabetes allows the effective management of the disease and reduces the risk of hypoglycemic or hyperglycemic episodes. Towards this goal, the development of reliable CGM models is essential for representing the corresponding signals and interpreting them with respect to factors and outcomes of interest. We propose a sparse decomposition model to approximate CGM time-series as a linear combination of a small set of exemplar atoms, appropriately designed through parametric functions to capture the main fluctuations of the CGM signal. Sparse decomposition is performed through the orthogonal matching pursuit (OMP). Results indicate that the proposed model provides 0.1 relative reconstruction error with 0.8 compression rate on a publicly available dataset containing 25 patients diagnosed with Type 1 diabetes. The atoms selected from the OMP procedure can be further interpreted in relation to the clinically meaningful components of the CGM signal (e.g. glucose spikes, hypoglycemic episodes, etc.

Plasma Insulin Levels and Hypoglycemia Affect Subcutaneous Interstitial Glucose Concentration.

Continuous glucose monitoring (CGM) accuracy during hypoglycemia is suboptimal. This might be partly explained by insulin or hypoglycemia-induced changes in the plasma interstitial subcutaneous (SC) fluid glucose gradient. The aim of the present study was to assess the role of plasma insulin (PI) and hypoglycemia itself in the plasma and interstitial SC fluid glucose concentration in patients with type 1 diabetes mellitus. Eleven subjects with type 1 diabetes (age 36.5 ± 9.1 years, HbA1c 7.9 ± 0.4% [62.8 ± 2.02 mmol/mol]; mean ± standard deviation) were evaluated under hyperinsulinemic euglycemia and hypoglycemia. Each subject underwent two randomized crossover clamps with either a primed 0.3 (low insulin) or 1 mU/(kg·min) (high insulin) insulin infusion. The raw CGM signal was normalized with median preclamp values to obtain a standardized measure of the interstitial glucose (IG) concentration before statistical analysis. The mean PI concentration was greater in high insulin studies (HISs) versus low insulin studies (LISs) (412.89 ± 13.63 vs. 177.22 ± 10.05 pmol/L). During hypoglycemia, glucagon, adrenaline, free fatty acids, glycerol, and beta-OH-butyrate were higher in the LIS (P < 0.0001). Likewise, the IG concentration was significantly different (P < 0.0001). This was due to lower IG concentration than plasma glucose (PG) concentration during the euglycemic hyperinsulinemic phases in the HIS. In contrast, no difference was observed during hypoglycemia. This was the result of an unchanged PG/IG gradient during the entire LIS, while in the HIS, this gradient increased during the hyperinsulinemic euglycemia phase. Both PI levels and hypoglycemia affect the relationship between IG and PG concentration. ClinicalTrials.gov Identifier: NCT01714895.

Multi-level supervision and modification of artificial pancreas control system

Artificial pancreas (AP) systems provide automated regulation of blood glucose concentration (BGC) for people with type 1 diabetes (T1D). An AP includes three components: a continuous glucose monitoring (CGM) sensor, a controller calculating insulin infusion rate based on the CGM signal, and a pump delivering the insulin amount calculated by the controller to the patient. The performance of the AP system depends on successful operation of these three components.Many APs use model predictive controllers that rely on models to predict BGC and to calculate the optimal insulin infusion rate. The performance of model-based controllers depends on the accuracy of the models that is affected by large dynamic changes in glucose-insulin metabolism or equipment performance that may move the operating conditions away from those used in developing the models and designing the control system. Sensor errors and missing signals will cause calculation of erroneous insulin infusion rates. And the performance of the controller may vary at each sampling step and each period (meal, exercise, and sleep), and from day to day.Here we describe a multi-level supervision and controller modification (ML-SCM) module is developed to supervise the performance of the AP system and retune the controller. It supervises AP performance in 3 time windows: sample level, period level, and day level. At sample level, an online controller performance assessment sub-module will generate controller performance indexes to evaluate various components of the AP system and conservatively modify the controller. A sensor error detection and signal reconciliation module will detect sensor error and reconcile the CGM sensor signal at each sample. At period level, the controller performance is evaluated with information collected during a certain time period and the controller is tuned more aggressively. At the day level, the daily CGM ranges are further analyzed to determine the adjustable range of controller parameters used for sample level and period level.Thirty subjects in the UVa/Padova metabolic simulator were used to evaluate the performance of the ML-SCM module and one clinical experiment is used to illustrate its performance in a clinical environment. The results indicate that the AP system with an ML-SCM module has a safer range of glucose concentration distribution and more appropriate insulin infusion rate suggestions than an AP system without the ML-SCM module.

An automatic glucose monitoring signal denoising method with noise level estimation and responsive filter updating

Abstract Although continuous glucose monitoring (CGM) devices have been the crucial part of the artificial pancreas, their success has been discounted by random measurement noise. The difficulty of denoising methods for CGM is that the filter parameters are hard to be determined to well reflect the internal blood glucose dynamics and the real noise level. Besides, the noise level may change from device to device, subject to subject and also within the subject as time goes on which thus requires that the filter parameters should be adjusted to follow the noise changes. In this paper, we proposed an automatic CGM signal denoising method which covers three important components. First, the state transition matrix which reveals the internal blood glucose dynamics and plays an important role in determining filter parameters can be estimated in response to different patients. Second, the real noise level can be estimated which are used to set the values of filter parameters properly. Third, a responsive filter updating rule is developed which can judge whether the values of filter parameters should be updated in response to the variability of signal-to-noise ratio. The process of dealing with the CGM signals is executing as follows: the model parameters and the noise level are evaluated using Expectation Maximization (EM) algorithm which can fix proper filter parameters for the current signals. Then, a confidence interval is defined by computing the power spectral density (PSD) of the CGM signals to identify the changes of noise level which can tell whether or not the parameters of Kalman filter (KF) should be adjusted. The above issues are investigated based on thirty in silico subjects and ten clinical subjects. The proposed method can work well to identify the changes of noise level and determine proper filter parameters.

A deep learning approach to adherence detection for type 2 diabetics.

Diabetes has become one of the biggest health problems in the world. In this context, adherence to insulin treatment is essential in order to avoid life-threatening complications. In this pilot study, a novel adherence detection algorithm using Deep Learning (DL) approaches was developed for type 2 diabetes (T2D) patients, based on simulated Continuous Glucose Monitoring (CGM) signals. A large and diverse amount of CGM signals were simulated for T2D patients using a T2D adapted version of the Medtronic Virtual Patient (MVP) model for T1D. By using these signals, different classification algorithms were compared using a comprehensive grid search. We contrast a standard logistic regression baseline to Multi- Layer Perceptrons (MLPs) and Convolutional Neural Networks (CNNs). The best classification performance with an average accuracy of 77:5% was achieved with CNN. Hence, this indicates the potential of DL, when considering adherence detection systems for T2D patients.

Reduction of Blood Glucose Measurements to Calibrate Subcutaneous Glucose Sensors: A Bayesian Multiday Framework.

In most continuous glucose monitoring (CGM) devices used for diabetes management, the electrical signal measured by the sensor is transformed to glucose concentration by a calibration function whose parameters are estimated using self-monitoring of blood glucose (SMBG) samples. The calibration function is usually a linear model approximating the nonlinear relationship between electrical signal and glucose concentration in certain time intervals. Thus, CGM devices require frequent calibrations, usually twice a day. The aim here is to develop a new method able to reduce the frequency of calibrations. The algorithm is based on a multiple-day model of sensor time-variability with second-order statistical priors on its unknown parameters. In an online setting, these parameters are numerically determined by the Bayesian estimation exploiting SMBG sparsely collected by the patient. The method is assessed retrospectively on 108 CGM signals acquired for 7 days by the Dexcom G4 Platinum sensor, testing progressively less-calibration scenarios. Despite the reduction of calibration frequency (on average from 2/day to 0.25/day), the method shows a statistically significant accuracy improvement compared to manufacturer calibration, e.g., mean absolute relative difference when compared to a laboratory reference decreases from 12.83% to 11.62% (p-value of 0.006). The methodology maintains (sometimes improves) CGM sensor accuracy compared to that of the original manufacturer, while reducing the frequency of calibrations. Reducing the need of calibrations facilitates the adoption of CGM technology both in terms of ease of use and cost, an obvious prerequisite for its use as replacement of traditional SMBG devices.

Hybrid online sensor error detection and functional redundancy for systems with time-varying parameters

Supervision and control systems rely on signals from sensors to receive information to monitor the operation of a system and adjust manipulated variables to achieve the control objective. However, sensor performance is often limited by their working conditions and sensors may also be subjected to interference by other devices. Many different types of sensor errors such as outliers, missing values, drifts and corruption with noise may occur during process operation. A hybrid online sensor error detection and functional redundancy system is developed to detect errors in online signals, and replace erroneous or missing values detected with model-based estimates. The proposed hybrid system relies on two techniques, an outlier-robust Kalman filter (ORKF) and a locally-weighted partial least squares (LW-PLS) regression model, which leverage the advantages of automatic measurement error elimination with ORKF and data-driven prediction with LW-PLS. The system includes a nominal angle analysis (NAA) method to distinguish between signal faults and large changes in sensor values caused by real dynamic changes in process operation. The performance of the system is illustrated with clinical data continuous glucose monitoring (CGM) sensors from people with type 1 diabetes. More than 50,000 CGM sensor errors were added to original CGM signals from 25 clinical experiments, then the performance of error detection and functional redundancy algorithms were analyzed. The results indicate that the proposed system can successfully detect most of the erroneous signals and substitute them with reasonable estimated values computed by functional redundancy system.

Continuous Glucose Monitoring Enables the Detection of Losses in Infusion Set Actuation (LISAs).

Reliable continuous glucose monitoring (CGM) enables a variety of advanced technology for the treatment of type 1 diabetes. In addition to artificial pancreas algorithms that use CGM to automate continuous subcutaneous insulin infusion (CSII), CGM can also inform fault detection algorithms that alert patients to problems in CGM or CSII. Losses in infusion set actuation (LISAs) can adversely affect clinical outcomes, resulting in hyperglycemia due to impaired insulin delivery. Prolonged hyperglycemia may lead to diabetic ketoacidosis—a serious metabolic complication in type 1 diabetes. Therefore, an algorithm for the detection of LISAs based on CGM and CSII signals was developed to improve patient safety. The LISA detection algorithm is trained retrospectively on data from 62 infusion set insertions from 20 patients. The algorithm collects glucose and insulin data, and computes relevant fault metrics over two different sliding windows; an alarm sounds when these fault metrics are exceeded. With the chosen algorithm parameters, the LISA detection strategy achieved a sensitivity of 71.8% and issued 0.28 false positives per day on the training data. Validation on two independent data sets confirmed that similar performance is seen on data that was not used for training. The developed algorithm is able to effectively alert patients to possible infusion set failures in open-loop scenarios, with limited evidence of its extension to closed-loop scenarios.

Exploring the Frequency Domain of Continuous Glucose Monitoring Signals to Improve Characterization of Glucose Variability and of Diabetic Profiles.

Continuous glucose monitoring (CGM) devices measure interstitial glucose concentrations (normally every 5 minutes), allowing observation of glucose variability (GV) patterns during the whole day. This information could be used to improve prescription of treatments and of insulin dosages for people suffering diabetes. Previous efforts have been focused on proposing indices of GV either in time or glucose domains, while the frequency domain has been explored only partially. The aim of this work is to explore the CGM signal in the frequency domain to understand if new indexes or features could be identified and contribute to a better characterization of glucose variability. The direct fast Fourier transform (FFT) and the Welch method were used to analyze CGM signals from three different profiles: people at risk of developing type 2 diabetes (P@R), T2D patients, and type 1 diabetes (T1D) patients. The results suggests that features extracted from the FFT (ie, the localization and power of the maximum peak of the power spectrum and the bandwidth at 3 dB) are able to provide a characterization for all the three populations under study compared with the Welch approach. Such preliminary results can represent a good insight for futures investigations with the possibility of building and using new indexes of glucose variability based on the frequency features.

From Two to One Per Day Calibration of Dexcom G4 Platinum by a Time-Varying Day-Specific Bayesian Prior.

In the DexCom G4 Platinum (DG4P) continuous glucose monitoring (CGM) sensor, the raw current signal generated by glucose-oxidase is transformed to glucose concentration by a calibration function whose parameters are periodically updated by matching self-monitoring of blood glucose references, usually twice a day, to compensate for sensor variability in time. The aim of this work is to reduce DG4P calibration frequency to once a day by a recently proposed Bayesian calibration algorithm, which employs a time-varying calibration function and suitable day-specific priors. The database consists of 57 CGM signals that are collected by the DG4P for 7 days. The Bayesian calibration algorithm is used to calibrate the raw current signal following two different schedules, that is, two and one calibration per day. Calibrated glycemic profiles are compared with those originally acquired by the manufacturer, on days 1, 4, and 7, where frequent blood glucose references were available, by using standard metrics, that is, mean absolute relative difference (MARD), percentage of accurate glucose estimates, and percentage of data in the A-zone of Clarke Error Grid. The one per day Bayesian calibration algorithm has accuracy similar to that of two per day (11.8% vs. 11.7% MARD, respectively), and it is statistically better (P-value of 0.0411) than the manufacturer calibration algorithm, which requires two calibrations per day (13.1% MARD). A Bayesian calibration algorithm employing a time-varying calibration function and suitable priors enables a reduction of the calibrations of DG4P sensor from two to one per day.