Surveillance Error Research Articles

Potential of Near-Infrared Optical Techniques for Non-invasive Blood Glucose Measurement: A Pilot Study

BackgroundDiabetes mellitus manifested by escalated blood glucose, demands periodic or continuous glucose monitoring in the pursuit of improved control, optimal management, and appropriate medication. The existing state-of-the-art for blood glucose estimation are invasive methodologies necessitating the acquisition of blood sample through either a finger-stick or venipuncture. MethodsWe propose a dual Near-infrared wavelength integrated in a 3D printed enclosure as the optical prototype for non-invasive glucose estimation. We acquired data from both diabetic and healthy subjects with the developed system and subsequently validated the system. ResultsThe system demonstrated commendable clinical accuracy, as evidenced by the alignment of data pairs representing actual blood glucose levels and blood glucose levels predicted by our optical system within the A+B zones of the Parkes error grid and zones of no-risk and lower risk as defined by the surveillance error grid. We achieved compliant pairs percentage of 95.6%, which satisfies the accuracy requirements of the blood glucose monitoring surveillance study. The mean absolute percentage error attained with the proposed device (5.99%) was significant in predicting the blood glucose. ConclusionWe successfully deployed the NIR wavelengths functionality as the promising approach for glucose monitoring, offering new possibilities for improved medical interventions.

The Diabetes Technology Society Error Grid and Trend Accuracy Matrix for Glucose Monitors.

An error grid compares measured versus reference glucose concentrations to assign clinical risk values to observed errors. Widely used error grids for blood glucose monitors (BGMs) have limited value because they do not also reflect clinical accuracy of continuous glucose monitors (CGMs). Diabetes Technology Society (DTS) convened 89 international experts in glucose monitoring to (1) smooth the borders of the Surveillance Error Grid (SEG) zones and create a user-friendly tool-the DTS Error Grid; (2) define five risk zones of clinical point accuracy (A-E) to be identical for BGMs and CGMs; (3) determine a relationship between DTS Error Grid percent in Zone A and mean absolute relative difference (MARD) from analyzing 22 BGM and nine CGM accuracy studies; and (4) create trend risk categories (1-5) for CGM trend accuracy. The DTS Error Grid for point accuracy contains five risk zones (A-E) with straight-line borders that can be applied to both BGM and CGM accuracy data. In a data set combining point accuracy data from 18 BGMs, 2.6% of total data pairs equally moved from Zones A to B and vice versa (SEG compared with DTS Error Grid). For every 1% increase in percent data in Zone A, the MARD decreased by approximately 0.33%. We also created a DTS Trend Accuracy Matrix with five trend risk categories (1-5) for CGM-reported trend indicators compared with reference trends calculated from reference glucose. The DTS Error Grid combines contemporary clinician input regarding clinical point accuracy for BGMs and CGMs. The DTS Trend Accuracy Matrix assesses accuracy of CGM trend indicators.

Continuous Glucose Monitoring and Recreational Scuba Diving in Type 1 Diabetes: Head-to-Head Comparison Between Free Style Libre 3 and Dexcom G7 Performance.

Background: Scuba diving was previously excluded because of hypoglycemic risks for patients with type 1 diabetes mellitus(T1DM). Specific eligibility criteria and a safety protocol have been defined, whereas continuous glucose monitoring (CGM) systems have enhanced diabetes management. This study aims to assess the feasibility and accuracy of CGM Dexcom G7 and Free Style Libre 3 in a setting of repetitive scuba diving in T1DM, exploring the possibility of nonadjunctive use. Material and Methods: The study was conducted during an event of Diabete Sommerso® association in 2023. Participants followed a safety protocol, with capillary glucose as reference standard (Beurer GL50Evo). Sensors' accuracy was evaluated through median and mean absolute relative difference (MeARD, MARD) and surveillance error grid (SEG). Data distribution and correlation were estimated by Spearman test and Bland-Altman plots. The ability of sensors to identify hypoglycemia was assessed by contingency tables. Results: Data from 202 dives of 13 patients were collected. The overall MARD was 31% (Dexcom G7) and 14.2% (Free Style Libre 3) and MeARD was 19.7% and 11.6%, respectively. Free Style Libre 3 exhibited better accuracy in normoglycemic and hyperglycemic ranges. SEG analysis showed 82.1% (Dexcom G7) and 97.4% (Free Style Libre 3) data on no-risk zone. Free Style Libre 3 better performed on hypoglycemia identification (diagnostic odds ratio of 254.10 vs. 58.95). Neither of the sensors reached the MARD for nonadjunctive use. Conclusions: The study reveals Free Style Libre 3 superior accuracy compared with Dexcom G7 in a setting of repetitive scuba diving in T1DM, except for hypoglycemic range. Both sensors fail to achieve accuracy for nonadjunctive use. Capillary tests remain crucial for safe dive planning, and sensor data should be interpreted cautiously. We suggest exploring additional factors potentially influencing sensor performance.

Using Artificial Intelligence to Improve the Accuracy of a Wrist-Worn, Noninvasive Glucose Monitor: A Pilot Study.

Self-monitoring of glucose is important to the successful management of diabetes; however, existing monitoring methods require a degree of invasive measurement which can be unpleasant for users. This study investigates the accuracy of a noninvasive glucose monitoring system that analyses spectral variations in microwave signals. An open-label, pilot design study was conducted with four cohorts (N = 5/cohort). In each session, a dial-resonating sensor (DRS) attached to the wrist automatically collected data every 60 seconds, with a novel artificial intelligence (AI) model converting signal resonance output to a glucose prediction. Plasma glucose was measured in venous blood samples every 5 minutes for Cohorts 1 to 3 and every 10 minutes for Cohort 4. Accuracy was evaluated by calculating the mean absolute relative difference (MARD) between the DRS and plasma glucose values. Accurate plasma glucose predictions were obtained across all four cohorts using a random sampling procedure applied to the full four-cohort data set, with an average MARD of 10.3%. A statistical analysis demonstrates the quality of these predictions, with a surveillance error grid (SEG) plot indicating no data pairs falling into the high-risk zones. These findings show that MARD values approaching accuracies comparable to current commercial alternatives can be obtained from a multiparticipant pilot study with the application of AI. Microwave biosensors and AI models show promise for improving the accuracy and convenience of glucose monitoring systems for people with diabetes.

Non-invasive Blood Glucose Monitoring Using a Radiofrequency Sensor and a Machine Learning Model

Diabetes Mellitus (DM) affects over 530 million people globally and is a condition characterized by high blood glucose (BG) that can result in severe long-term health consequences if poorly managed. To mitigate the risks associated with DM, it is crucial to regulate BG levels through regular monitoring. Current methods of monitoring BG come with drawbacks in the form of pain and expense. With the rapid increase in prevalence of DM, there is an increasing need for economical, accurate, and non-invasive continuous measures of BG. Here we present a validation for a novel sensor that rapidly scans through a wide range of radio frequencies (RF) and uses machine learning (ML) techniques for the purpose of non-invasive BG measurement. After model training, the RF sensor and ML techniques employed in this study can predict BG at a level significantly better than chance, as shown below. Data were collected from 13 healthy participants over a series of tests lasting 2 – 3 hours. During each test, participants placed their forearm on an RF sensor that measured their BG levels using sweeps across the 500 MHz – 1500 MHz range at 0.1 MHz intervals. Participants ingested 37.5 grams of glucose solution to generate BG readings across normoglycemic and hyperglycemic ranges. Each sweep took approximately 22 seconds, including a one second pause between sweeps. Across 110 tests, we collected 3,311 BG observations. Concurrent measurements from a Continuous Glucose Monitor (CGM) were taken as reference. Data were divided using a 60-20-20 (training-validation-test) split and trained on a Light Gradient-Boosting Machine (lightGBM) model to predict BG values. A ‘blind’ evaluation of model performance was determined using the held-out test dataset. We predicted BG values in the held-out test dataset with a Mean Absolute Relative Difference (MARD) of 10.8% in the normoglycemic range and 15.9% in the hyperglycemic range (11.27% overall MARD). These results were significantly better than a chance model (two-sample t-test, p <0.01). In a Surveillance Error Grid analysis of model accuracy, 89.0% of predictions fell in Risk Level 0, 10.1% in Risk Level 1, and 0.9% in Risk Level 2. These results demonstrate that the novel RF sensor and ML techniques described can predict a reference BG value in the population under study. More research is underway to refine and expand these methods using a gold-standard blood glucose reference device and among a broader participant population with an expanded glucose range. DK and VS are consultants for and own stock in Know Labs. JA and KC are employed by and have stock options in Know Labs. CW and KP are consultants for Know Labs. NA. This is the full abstract presented at the American Physiology Summit 2024 meeting and is only available in HTML format. There are no additional versions or additional content available for this abstract. Physiology was not involved in the peer review process.

Non-invasive glucose prediction and classification using NIR technology with machine learning

In this paper, a dual wavelength short near-infrared system is described for the detection of glucose levels. The system aims to improve the accuracy of blood glucose detection in a cost-effective and non-invasive way. The accuracy of the method is evaluated using real-time samples collected with the reference finger prick glucose device. A feed forward neural network (FFNN) regression method is employed to predict glucose levels based on the input data obtained from NIR technology. The system calculates glucose evaluation metrics and performs Surveillance error grid (SEG) analysis. The coefficient of determination R2 and mean absolute error are observed 0.99 and 2.49 mg/dl, respectively. Additionally, the system determines the root mean square error (RMSE) as 3.02 mg/dl. It also shows that the mean absolute percentage error (MAPE) is 1.94% and mean squared error (MSE) is 9.16 (mg/dl)2 for FFNN. The SEG analysis shows that the glucose values measured by the system fall within the clinically acceptable range when compared to the reference method. Finally, the system uses the multi-class classification method of the multilayer perceptron (MLP) and K-nearest neighbors (KNN) classifier to classify glucose levels with an accuracy of 99%.

Accuracy and Feasibility of Real-time Continuous Glucose Monitoring in Critically Ill Patients After Abdominal Surgery and Solid Organ Transplantation.

Glycemia management in critical care is posing a challenge in frequent measuring and adequate insulin dose adjustment. In recent years, continuous glucose measurement has gained accuracy and reliability in outpatient and inpatient settings. The aim of this study was to assess the feasibility and accuracy of real-time continuous glucose monitoring (CGM) in ICU patients after major abdominal surgery. We included patients undergoing pancreatic surgery and solid organ transplantation (liver, pancreas, islets of Langerhans, kidney) requiring an ICU stay after surgery. We used a Dexcom G6 sensor, placed in the infraclavicular region, for real-time CGM. Arterial blood glucose measured by the amperometric principle (ABL 800; Radiometer, Copenhagen, Denmark) served as a reference value and for calibration. Blood glucose was also routinely monitored by a StatStrip bedside glucose meter. Sensor accuracy was assessed by mean absolute relative difference (MARD), bias, modified Bland-Altman plot, and surveillance error grid for paired samples of glucose values from CGM and acid-base analyzer (ABL). We analyzed data from 61 patients and obtained 1,546 paired glucose values from CGM and ABL. Active sensor use was 95.1%. MARD was 9.4%, relative bias was 1.4%, and 92.8% of values fell in zone A, 6.1% fell in zone B, and 1.2% fell in zone C of the surveillance error grid. Median time in range was 78%, with minimum (<1%) time spent in hypoglycemia. StatStrip glucose meter MARD compared with ABL was 5.8%. Our study shows clinically applicable accuracy and reliability of Dexcom G6 CGM in postoperative ICU patients and a feasible alternative sensor placement site.

Accuracy of a Non-Invasive Home Glucose Monitor for Measurement of Blood Glucose.

Patients with diabetes mellitus monitor their blood glucose at home with monitors that require a drop of blood or use a continuous glucose monitoring device that implants a small needle in the body. However, both cause discomfort to the patients which may inhibit them for regular blood glucose checks. Photoplethysmogram (PPG) sensing technology is an approach for non-invasive blood glucose measurement and PPG sensors can be used to predict hypoglycaemic episodes. InChcek is a PPG-based non-invasive glucose monitor. However, its accuracy has not been checked yet. Hence, this study aimed to evaluate the accuracy of InCheck, a non-invasive glucose monitor for the estimation of blood glucose. In a tertiary care hospital, patients who came for blood glucose estimation were tested for blood glucose non-invasively on the InCheck device and then by the laboratory method (glucose oxidase-peroxidase). These two readings were compared. We used International Organization for Standardization (ISO) 15197:2013 (95% of values should be within ± 15 mg/dL of reference reading if reference glucose <100 mg/dL or within ± 15% of reference reading if reference glucose ≥100 mg/dL and 99% of the values should be within zones A and B in consensus error grid), and Surveillance Error Grid for analyzing the accuracy. A total of 1223 samples were analyzed. There was a significant difference between the reference method glucose level (135 [Q1-Q3: 97 - 179] mg/dL) and monitor-measured glucose level (188.33 [Q1-Q3: 167.33-209.33] mg/dL) (P < 0.0001). A total of 18.5% of readings were following ISO 15197:2013 criteria and 67.25% of coordinates were within zone A and zone B of the consensus error grid. In the surveillance error grid analysis, about 29.4% of values were in the no-risk zone, 51.8% in slight risk, 18.6% in moderate risk, and 0.2% were in the severe risk zone. The accuracy of the InCheck device for the estimation of blood glucose by PPG signal is not following the recommended guidelines. Hence, further research is necessary for programming or redesigning the hardware and software for a better result from this optical sensor-based non-invasive home glucose monitor.

Evaluation of a hybrid protocol using continuous glucose monitoring and point-of-care testing in non-critically ill patients in a community hospital.

Inpatient glycemic management typically involves use of point-of-care (POC) glucose measurements to inform insulin dosing decisions. This study evaluated a hybrid monitoring protocol using real-time continuous glucose monitoring (rtCGM) supplemented with POC testing at a community hospital. Adult inpatients receiving POC glucose testing were monitored using rtCGM in a telemetry unit. The hybrid monitoring protocol required a once-daily POC test but otherwise primarily relied on rtCGM values for insulin dosing decisions. Outcomes assessment included surveillance error grid (SEG) and Clarke Error Grid (CEG) analysis results, the mean absolute relative difference (MARD) for available rtCGM-POC value pairs before and after study protocol application, the number of POC tests avoided, and the number of hypoglycemic events involving a blood glucose value of <70 mg/dL identified by rtCGM and POC values. Data were collected from 30 inpatients (the mean age was 69.4 years, 77% were female, 80% had type 2 diabetes, and 37% were at-home insulin users). With the protocol applied, a total of 202 rtCGM-POC pairs produced a MARD of 12.5%. SEG analysis showed 2 pairs in the "moderate" risk category, with all other pairs in the "none" or "slight" risk categories. CEG analysis showed 99% of paired values to be in the clinically acceptable range. Six hypoglycemic events in 5 patients were resolved without incident. Three hundred three POC tests were avoided, a 60% reduction for the study duration. Use of a hybrid monitoring protocol of rtCGM and POC testing in a community hospital demonstrated sustained rtCGM accuracy and was found to reduce the frequency of POC testing to manage inpatient glycemia.

Flash glucose monitoring system in gestational diabetes: a study of accuracy and usability.

Studies of flash glucose monitoring systems (FGMSs) in pregnancy are insufficient, especially in gestational diabetes (GD). Our aim was to evaluate Freestyle Libre's usability and accuracy (compared to self-monitoring of blood glucose [SMBG]) for GD patients in real-life conditions. This is a prospective study with pregnant women diagnosed with GD (n = 24 for the usability analysis; n = 19 for the accuracy analysis). The study duration was up to 28 days (lifetime of two sensors). Participants executed a minimum of four daily FGMS readings obtained immediately after capillary SMBG. Analytical accuracy was assessed with mean absolute relative difference (MARD) and mean absolute difference (MAD); clinical accuracy was assessed with Surveillance Error Grid (SEG). Usability was evaluated with a user acceptability questionnaire. The mean pregestational BMI was 25.21 ± 5.15 kg/m2 (mean ± SD), the mean gestational age was 30.31 ± 2.02 weeks, and the mean glucose values were 76.63 ± 7.49 mg/dL. A total of 1339 SMBG-FGMS pairs of values were obtained. Analytical accuracy was good with an overall MARD of 14.07% and an in-target MARD of 13.79%. The number of SMBG-FMGS pairs for above-target values was low (122 of 1339) with an associated MARD of 17.95%. Clinical accuracy of the FGMS was demonstrated, with 94.4% of values in the no-risk or slight, lower risk zones of the SEG. FGMS accuracy was unaffected by pregestational BMI or gestational age. The user acceptability questionnaire showed high levels of satisfaction, with 95.8-100% preferring FGMS to SMBG. No unexpected or severe adverse effects occurred. FGMS showed good performance in GD regarding accuracy and usability. Larger studies are needed to corroborate our results, verify the analytical accuracy of above-target values as this glucose range might lead to initiation or adjustment of pharmacological therapy, and ultimately establish definitive recommendations regarding prescription of FGMS for GD patients.

Clinical Use of Continuous Glucose Monitoring in Critically Ill Pediatric Patients with Diabetic Ketoacidosis.

Background: The use of continuous glucose monitoring (CGM) in pediatric patients with diabetic ketoacidosis (DKA) remains investigational, and data on its accuracy in pediatric intensive care units (PICU) are limited. This study evaluated the accuracy of three CGM devices in pediatric patients with DKA in the PICU. Methods: We compared 399 matched pairs of CGM and point-of-care capillary glucose (POC) values and grouped patients based on whether they changed their CGM sensor during their PICU stay. Results: Eighteen patients with a mean age of 10.98 ± 4.20 years were included, with three patients in the sensor change group. The overall mean absolute relative difference (MARD) was 13.02%. The Medtronic Guardian Sensor 3 (n = 331), Dexcom G6 (n = 41), and Abbott FreeStyle Libre 1 (n = 27) showed MARD values of 13.40%, 11.12%, and 11.33%, respectively. The surveillance error grid (SEG), Bland-Altman plot, and Pearson's correlation coefficient demonstrated satisfactory clinical accuracy of the CGM devices (SEG zones A and B, 98.5%; mean difference, 15.5 mg/dL; Pearson's correlation coefficient [r2], 0.76, P < 0.0001). MARD was significantly lower in subjects who did not experience a sensor change (11.74% vs. 17.31%, P = 0.048). Also, a statistically significant negative correlation was found between serum bicarbonate levels and POC-CGM values (r = -0.34, P < 0.001). Conclusions: The severity of DKA has a major effect on reducing the accuracy of the CGM, especially during the first several days in the intensive care unit. The reduced accuracy appears to be related to acidosis, as reflected in the serum bicarbonate levels.

228-OR: Accuracy and Reliability of Real-Time Continuous Glucose Monitoring in the Intensive Care Unit after Major Abdominal Surgery

Introduction: Blood glucose control in patients requiring intensive care remains a therapeutic challenge; however, recent advancements in glucose sensing technologies have sparked interest in the potential use of real-time continuous glucose monitoring in the subcutaneous compartment (rtCGM) also in the intensive care unit (ICU) setting. We performed a clinical trial assessing the accuracy and reliability of rtCGM in postoperative ICU patients after major abdominal surgery. Methods: Patients undergoing pancreatic surgery, liver transplantation, pancreas (or pancreatic islets) and kidney transplantation were enrolled in the trial. Dexcom G6 (Dexcom, Inc., San Diego, USA) was used for rtCGM. Arterial blood glucose measured by the amperometric principle (ABL 800, Radiometer, Copenhagen, Denmark) served as reference and to calibrate the Dexcom G6 (every 6 hours on day 1, once daily day 2 and 3). Sensor accuracy was assessed by computing mean absolute relative difference (MARD), bias, modified Bland-Altman plot and surveillance error grid for paired samples of glucose values from CGM and ABL. Results: Sixty-one patients after major abdominal surgery staying in ICU post-operatively were included into this analysis. Median monitoring length was 7 days with a 97.1 ± 6.5% proportion of active CGM use. Overall, 1566 paired glucose values measured with CGM and ABL were obtained. MARD for CGM compared with ABL-measured glucose was 9.7% with a bias of 0.9% and coefficient of variation of 14%. In the surveillance error grid analysis, 90.9% of pairs were in the 0 risk zone and 7.9% in the risk zone 1 (out of 7). Throughout the study 3 sensors needed to be replaced due to technical problems. Conclusion: Our results show a clinically applicable accuracy and reliability of the Dexcom G6 continuous glucose monitoring system in postoperative ICU. This opens up new possibilities in intensifying ICU glucose control, lowering hypoglycaemia risk and reducing nursing staff workload. Disclosure B.Hagerf (voglová): None. M.Protus: None. L.Nemetova: None. M.Mraz: None. M.Haluzik: Advisory Panel; Novo Nordisk, Lilly Diabetes, Boehringer-Ingelheim, Research Support; Sanofi, Speaker's Bureau; Abbott, AstraZeneca. P.Girman: None. J.Franekova: None. V.Svirlochova: None. A.Jabor: None. Funding Charles University; Czech Ministry of Health; Institute for Clinical and Experimental Medicine (IKEM, IN00023001); National Institute for Research of Metabolic and Cardiovascular Diseases (EXCELES, LX22NPO5104)

Diabetes Complications and Related Comorbidities Impair the Accuracy of FreeStyle Libre, a Flash Continuous Glucose Monitoring System, in Patients with Type 2 Diabetes.

Although flash continuous glucose monitoring systems (FCGM) accuracy has been extensively studied in diabetes, its accuracy is still not fully evaluated in type 2 diabetes (T2D) patients in real-world settings. In the present study, we aim to assess the effects of diabetes complications and related comorbidities on FCGM accuracy in T2D patients with diabetes complications and related comorbidities in the real world. FCGM data were collected at eight-time points daily (3 AM, 7 AM, 9 AM, 11 AM, 1 PM, 5 PM, 7 PM, and 9 PM) from 742 patients with T2D and compared with simultaneous fingertip capillary blood glucose (reference blood glucose, REF), and the difference was evaluated using Parkes error grid (PEG), surveillance error grid (SEG), and logistic regression analysis. In total, 25,579 FCGM/REF data pairs were included in the study. The FCGM values were lower than the paired REF values in 75% of the pairs. The maximum bias (-23.0%) and maximum mean absolute relative difference (24.5%) were observed at 3 AM among eight-time points. SEG analysis also demonstrated the highest percentage of paired readings in moderate and great risk zone (C and D) at 3 AM than PEG analysis (7.33% vs 0.43%, P<0.001). According to the SEG classification, hypoglycemia, infection, diabetic foot, diabetic ketoacidosis, and hypertension were independent risk factors that impaired FCGM accuracy in patients. FCGM commonly underestimates blood glucose levels. Compared with PEG, SEG analysis seems more conducive to the analysis of FCGM performance. The present data highlights the impairment of diabetes complications and related comorbidities on the FCGM accuracy in T2D patients.

Evaluation of the FreeStyle Libre, a flash glucose monitoring system, in client-owned cats with diabetes mellitus.

ObjectivesHome blood glucose monitoring using a portable blood glucose meter is important in the management of feline diabetes mellitus, but taking blood samples may be stressful for owners and cats. A flash glucose monitoring system measuring interstitial glucose, such as the FreeStyle Libre, overcomes some of these drawbacks. The aim of this study was to evaluate the practical use and analytical and clinical accuracy of the FreeStyle Libre in 41 client-owned diabetic cats.MethodsIn this prospective study, interstitial glucose concentrations were measured with the FreeStyle Libre and compared with blood glucose concentrations measured with a portable blood glucose meter (AlphaTRAK) on days 1, 7 or 8 and 14 after application of the device. Cat behaviour during application, location, skin reaction at the attachment site and owner satisfaction were assessed. Accuracy was determined by fulfilment of ISO 15197:2013 criteria, including Bland–Altman plotting and error grid analysis.ResultsPlacing the device was easy, with 70% of cats showing no reaction. Most sensors were placed on the thoracic wall. Skin reactions at the attachment site were not present or mild in almost all cats. Owners were very satisfied with the use of the FreeStyle Libre. Median functional life of the sensor was 10 days (range 1–14). Good correlation was found between interstitial and blood glucose measurements (rho[r] = 0.88, P <0.0001). Fifty-three percent of interstitial glucose concentrations were within a maximum deviation of 15% from blood glucose concentrations and 92.7% were within the safe risk zones 0 and 1 of the surveillance error grid.Conclusions and relevanceThe flash glucose monitoring system was easy to use and owners of diabetic cats were satisfied with its use. Although the device did not completely fulfil ISO requirements, it is sufficiently accurate for glucose monitoring in diabetic cats.

Investigational Outcomes of Normal and Diabetic Human Volunteers using Microwave based Non-invasive Blood Glucometer

—Noninvasive estimation of blood glucose using microwave based sensor is one of the most challenging research fields in recent days. These methods are more adaptable to avoid soreness and skin estrangement that occur in traditional invasive blood glucose measurement techniques. This paper suggests the investigational outcomes of a microwave based noninvasive blood glucose meter to check performance of the microwave antenna sensor. Narrowband micro strip antenna with resonating microwave frequency of 1.3 GHz is utilized to measure Blood Glucose Level (BGL) noninvasively. Validity of sensor is assessed by performing medical tests on non diabetic and diabetic human volunteers. Mean Absolute Relative Difference (MARD) and Surveillance Error Grid (SEG) analyses were performed during experimentation to find deviation of estimated BGL from reference BGL which is measured using available invasive technique. Direct assessment of estimated data with reference data provides excellent reliability and repeatability of proposed microwave technique. Keywords- glucometer, Blood Glucose Level (BGL), medical tests, Mean Absolute Relative Difference (MARD), Surveillance Error Grid (SEG)

Predicting glucose level with an adapted branch predictor

Background and objectiveDiabetes mellitus manifests as prolonged elevated blood glucose levels resulting from impaired insulin production. Such high glucose levels over a long period of time damage multiple internal organs. To mitigate this condition, researchers and engineers have developed the closed loop artificial pancreas consisting of a continuous glucose monitor and an insulin pump connected via a microcontroller or smartphone. A problem, however, is how to accurately predict short term future glucose levels in order to exert efficient glucose-level control. Much work in the literature focuses on least prediction error as a key metric and therefore pursues complex prediction methods such a deep learning. Such an approach neglects other important and significant design issues such as method complexity (impacting interpretability and safety), hardware requirements for low-power devices such as the insulin pump, the required amount of input data for training (potentially rendering the method infeasible for new patients), and the fact that very small improvements in accuracy may not have significant clinical benefit. MethodsWe propose a novel low-complexity, explainable blood glucose prediction method derived from the Intel P6 branch predictor algorithm. We use Meta-Differential Evolution to determine predictor parameters on training data splits of the benchmark datasets we use. A comparison is made between our new algorithm and a state-of-the-art deep-learning method for blood glucose level prediction. ResultsTo evaluate the new method, the Blood Glucose Level Prediction Challenge benchmark dataset is utilised. On the official test data split after training, the state-of-the-art deep learning method predicted glucose levels 30 min ahead of current time with 96.3% of predicted glucose levels having relative error less than 30% (which is equivalent to the safe zone of the Surveillance Error Grid). Our simpler, interpretable approach prolonged the prediction horizon by another 5 min with 95.8% of predicted glucose levels of all patients having relative error less than 30%. ConclusionsWhen considering predictive performance as assessed using the Blood Glucose Level Prediction Challenge benchmark dataset and Surveillance Error Grid metrics, we found that the new algorithm delivered comparable predictive accuracy performance, while operating only on the glucose-level signal with considerably less computational complexity.

Practical implementation of remote continuous glucose monitoring in hospitalized patients with diabetes.

PurposeInpatient diabetes management involves frequent assessment of glucose levels for treatment decisions. Here we describe a program for inpatient real-time continuous glucose monitoring (rtCGM) at a community hospital and the accuracy of rtCGM-based glucose estimates.MethodsAdult inpatients with preexisting diabetes managed with intensive insulin therapy and a diagnosis of coronavirus disease 2019 (COVID-19) were monitored via rtCGM for safety. An rtCGM system transmitted glucose concentration and trending information at 5-minute intervals to nearby smartphones, which relayed the data to a centralized monitoring station. Hypoglycemia alerts were triggered by rtCGM values of ≤85 mg/dL, but rtCGM data were otherwise not used in management decisions; insulin dosing adjustments were based on blood glucose values measured via fingerstick blood sampling. Accuracy was evaluated retrospectively by comparing rtCGM values to contemporaneous point-of-care (POC) blood glucose values.ResultsA total of 238 pairs of rtCGM and POC data points from 10 patients showed an overall mean absolute relative difference (MARD) of 10.3%. Clarke error grid analysis showed 99.2% of points in the clinically acceptable range, and surveillance error grid analysis showed 89.1% of points in the lowest risk category. It was determined that for 25% of the rtCGM values, discordances in rtCGM and POC values would likely have resulted in different insulin doses. Insulin dose recommendations based on rtCGM values differed by 1 to 3 units from POC-based recommendations.ConclusionrtCGM for inpatient diabetes monitoring is feasible. Evaluation of individual rtCGM-POC paired values suggested that using rtCGM data for management decisions poses minimal risks to patients. Further studies to establish the safety and cost implications of using rtCGM data for inpatient diabetes management decisions are warranted.

Post-Market Surveillance of a Blood Glucose Test Strip Demonstrates No Evidence of Interference on Clinical Accuracy in a Large Cohort of People with Type 1 or Type 2 Diabetes.

Regulations and industry guidance relating to testing for interference in blood glucose monitoring (BGM) systems continue to focus on in vitro laboratory bench tests. Post-market surveillance (PMS) in a clinical setting allows for BGM accuracy assessments to evaluate the impact of real-world exposure to polypharmacy in people with diabetes. This study evaluated the OneTouch Select Plus® BGM test-strip accuracy with respect to polypharmacy using a clinical registry dataset. Medication profiles were analysed for 1023 subjects (425 with type 1 (T1D) and 598 with type 2 diabetes (T2D)) attending 3 UK hospitals. Blood samples were analysed to determine clinical accuracy of the BGM test-strip against a laboratory comparator. 538 different medications (48 diabetes and 490 non-diabetes) were recorded across the 1023 subjects. Patients took on average 6.9 (n = 1-36) individual medications and 4.1 (n = 1-13) unique medication classes. Clinical accuracy to EN ISO 15197:2015 criteria were met irrespective of increasing average number of individual medications, categorized from 1-3, 4-6, 7-9, 10-12 and >12 taken per subject (97.7%, 97.7%, 97.8%, 97.8%, and 98.4%, respectively). Clinical accuracy criteria were met across 15 classes of medication using the combined dataset (97.9%; 29784/30433). Surveillance Error Grid (SEG) analysis showed 98.7% (29959/30368) of readings presented no clinical risk. No individual class or combination of medication classes impacted clinical accuracy of the BGM test-strip. Clinical performance for the test strip under assessment demonstrated no evidence of interference from over 500 prescription medications, with clinical accuracy maintained across a range of polypharmacy conditions in people with diabetes.

The Accuracy of Continuous Glucose Monitoring in the Medical Intensive Care Unit.

To assess the accuracy of continuous glucose monitoring (CGM) in medical intensive care unit (MICU) patients. A Medtronic Enlite® sensor accuracy was assessed versus capillary blood glucose (CBG) and plasma glucose (PG) using the mean absolute relative difference (MARD), surveillance error grid (SEG) analysis and modified Bland-Altman plots. Using CBG as a reference, MARD was 6.6%. Overall, 99.7% of the CGM readings were within the "no risk" zone. No significant differences in accuracy were seen within vasopressor subgroups. Using PG as the reference, MARD was 8.8%. The surveillance error grid analysis showed 95.2% of glucose readings were within the "no risk" zone. There were no device-related adverse events. The CGM sensor showed acceptable accuracy in MICU patients, regardless of vasopressor use.

Machine learning-based glucose prediction with use of continuous glucose and physical activity monitoring data: The Maastricht Study.

Closed-loop insulin delivery systems, which integrate continuous glucose monitoring (CGM) and algorithms that continuously guide insulin dosing, have been shown to improve glycaemic control. The ability to predict future glucose values can further optimize such devices. In this study, we used machine learning to train models in predicting future glucose levels based on prior CGM and accelerometry data. We used data from The Maastricht Study, an observational population-based cohort that comprises individuals with normal glucose metabolism, prediabetes, or type 2 diabetes. We included individuals who underwent >48h of CGM (n = 851), most of whom (n = 540) simultaneously wore an accelerometer to assess physical activity. A random subset of individuals was used to train models in predicting glucose levels at 15- and 60-minute intervals based on either CGM data or both CGM and accelerometer data. In the remaining individuals, model performance was evaluated with root-mean-square error (RMSE), Spearman's correlation coefficient (rho) and surveillance error grid. For a proof-of-concept translation, CGM-based prediction models were optimized and validated with the use of data from individuals with type 1 diabetes (OhioT1DM Dataset, n = 6). Models trained with CGM data were able to accurately predict glucose values at 15 (RMSE: 0.19mmol/L; rho: 0.96) and 60 minutes (RMSE: 0.59mmol/L, rho: 0.72). Model performance was comparable in individuals with type 2 diabetes. Incorporation of accelerometer data only slightly improved prediction. The error grid results indicated that model predictions were clinically safe (15 min: >99%, 60 min >98%). Our prediction models translated well to individuals with type 1 diabetes, which is reflected by high accuracy (RMSEs for 15 and 60 minutes of 0.43 and 1.73 mmol/L, respectively) and clinical safety (15 min: >99%, 60 min: >91%). Machine learning-based models are able to accurately and safely predict glucose values at 15- and 60-minute intervals based on CGM data only. Future research should further optimize the models for implementation in closed-loop insulin delivery systems.