Continuous Glucose Monitoring Readings Research Articles

Performance of subcutaneous continuous glucose monitoring in adult critically ill patients receiving vasopressor therapy.

Background Subcutaneous continuous glucose monitoring (CGM) may facilitate glucose control in the ICU. We aimed to assess the accuracy of CGM (Dexcom G6) against arterial blood glucose (ABG) in adult critically ill patients receiving intravenous insulin infusion and vasopressor therapy. We also aimed to assess feasibility and tolerability of CGM in this setting. Methods We included ICU patients receiving mechanical ventilation, insulin, and vasopressor therapy. Numerical accuracy was assessed by the mean absolute relative difference (MARD), overall, across arterial glucose strata, over different noradrenaline equivalent infusion rates, and over time since CGM start. MARD below 14% was considered acceptable. Clinical accuracy was assessed using Clarke Error Grid (CEG) analysis. Feasibility outcome included number and duration of interrupted sensor readings due to signal loss. Tolerability outcome included skin reactions related to sensor insertion or sensor adhesives. Results We obtained 2946 paired samples from 40 patients (18 with type 2 diabetes) receiving a median (IQR) maximum noradrenaline equivalent infusion rate of 0.18 (0.08-0.33) µg/kg/min during CGM. Overall, MARD was 12.7 (95% CI 10.7-15.3) % and 99.8% of CGM readings were within CEG zones A and B. MARD values ≥14% were observed when ABG was outside target range (6-10 mmol/l [108-180 mg/dl]) and with noradrenaline equivalent infusion rates above 0.10 µg/kg/min. Accuracy improved with time after CGM start, reaching MARD values below 14% after 36 hours. We observed four episodes of interrupted sensor readings due to signal loss ranging from 5 to 20 minutes. We observed no skin reactions related to sensor insertion or sensor adhesives. Conclusions In our ICU cohort of patients receiving vasopressor infusion, subcutaneous CGM demonstrated acceptable overall numerical and clinical accuracy. However, suboptimal accuracy may occur outside glucose ranges of 6-10 mmol/l (108-180 mg/dl), during higher dose vasopressor infusion, and during the first 36 hours after CGM start.

Tailored Meal-Type Food Provision for Diabetes Patients Can Improve Routine Blood Glucose Management in Patients with Type 2 Diabetes: A Crossover Study.

This study aimed to determine whether patients with type 2 diabetes can benefit from a meal plan designed based on diabetes management guidelines to improve blood glucose levels. Participants were divided into intervention and control groups. The intervention group consumed a diabetic diet for 2 weeks, while the control group consumed their normal diet. After 2 weeks, the groups switched their dietary regimens. The participants' demographic and clinical characteristics were evaluated, including factors such as blood pressure, blood lipid levels, weight and waist circumference, blood glucose levels (self-monitored and continuously monitored), nutritional status, and blood-based markers of nutrient intake. The dietary intervention group improved waist circumference, body fat percentage, low-density lipoprotein cholesterol, triglyceride levels, and glucose. The energy composition ratio of carbohydrates and proteins changed favorably, and sugar intake decreased. In addition, the proportion of continuous glucose monitoring readings within the range of 180-250 mg/dL was relatively lower in the intervention group than that of the control group. Meals designed based on diabetes management guidelines can improve clinical factors, including stable blood glucose levels in daily life, significantly decrease the carbohydrate energy ratio, and increase the protein energy ratio. This study can help determine the role of dietary interventions in diabetes management and outcomes.

The Discrepancy Between Hemoglobin A1c and Glucose Management Indicators in 26 Patients Treated With Continuous Glucose Monitoring in an Internal Medicine Residency Clinic.

We conducted a retrospective observational cohort study between 2020 and 2023 in 26 patients with type 1 and type 2 diabetes mellitus (DM) who were using 3-4 injections per day of insulin and were monitored by continuous glucose monitoring (CGM). The goal of this retrospective observational cohort study is to compare these two metrics in an internal medicine community primary care residency clinic. We used CGM devices, Dexcom G6 and G7, and Freestyle Libre 3. The goal was to compare the patient's hemoglobin A1c (HbA1c) taken during their clinic visit by phlebotomy as a marker for diabetic control with an estimated HbA1c glucose management indicator (GMI) derived from the 30-day CGM readings. HbA1c is derived from the blood, while the GMI value is derived from the interstitial fluid. Both parameters were taken within 30 days of each other. GMI was taken in the last 30 days. We excluded patients with known anemia, chronic kidney disease, polycythemia, cirrhosis of the liver, or metabolic dysfunction associated with steatohepatitis (MASH) because disease states can affect the measured HbA1c. Also, pregnantand African American patients were excluded. We concluded the measured HbA1c was 0.34% (4 mmol/mol) higher than the CGM-derived GMI. The relationship between factors that affect glycemic control was discussed in the article, as well as the future utilization of them in improving diabetic control and management. As the use of CGM continues to grow, addressing differences between laboratory-measured HbA1c and CGM-derived GMI is critical.

A Machine Learning Model for Week-Ahead Hypoglycemia Prediction From Continuous Glucose Monitoring Data.

Remote patient monitoring (RPM) programs augment type 1 diabetes (T1D) care based on retrospective continuous glucose monitoring (CGM) data. Few methods are available to estimate the likelihood of a patient experiencing clinically significant hypoglycemia within one week. We developed a machine learning model to estimate the probability that a patient will experience a clinically significant hypoglycemic event, defined as CGM readings below 54 mg/dL for at least 15 consecutive minutes, within one week. The model takes as input the patient's CGM time series over a given week, and outputs the predicted probability of a clinically significant hypoglycemic event the following week. We used 10-fold cross-validation and external validation (testing on cohorts different from the training cohort) to evaluate performance. We used CGM data from three different cohorts of patients with T1D: REPLACE-BG (226 patients), Juvenile Diabetes Research Foundation (JDRF; 355 patients) and Tidepool (120 patients). In 10-fold cross-validation, the average area under the receiver operating characteristic curve (ROC-AUC) was 0.77 (standard deviation [SD]: 0.0233) on the REPLACE-BG cohort, 0.74 (SD: 0.0188) on the JDRF cohort, and 0.76 (SD: 0.02) on the Tidepool cohort. In external validation, the average ROC-AUC across the three cohorts was 0.74 (SD: 0.0262). We developed a machine learning algorithm to estimate the probability of a clinically significant hypoglycemic event within one week. Predictive algorithms may provide diabetes care providers using RPM with additional context when prioritizing T1D patients for review.

Analytical Performance of the Factory-Calibrated Flash Glucose Monitoring System FreeStyle Libre2TM in Healthy Women.

Continuous glucose monitoring (CGM) is used clinically and for research purposes to capture glycaemic profiles. The accuracy of CGM among healthy populations has not been widely assessed. This study assessed agreement between glucose concentrations obtained from venous plasma and from CGM (FreeStyle Libre2TM, Abbott Diabetes Care, Witney, UK) in healthy women. Glucose concentrations were assessed after fasting and every 15 min after a standardized breakfast over a 4-h lab period. Accuracy of CGM was determined by Bland-Altman plot, 15/15% sensor agreement analysis, Clarke error grid analysis (EGA) and mean absolute relative difference (MARD). In all, 429 valid CGM readings with paired venous plasma glucose (VPG) values were obtained from 29 healthy women. Mean CGM readings were 1.14 mmol/L (95% CI: 0.97 to 1.30 mmol/L, p < 0.001) higher than VPG concentrations. Ratio 95% limits of agreement were from 0.68 to 2.20, and a proportional bias (slope: 0.22) was reported. Additionally, 45% of the CGM readings were within ±0.83 mmol/L (±15 mg/dL) or ±15% of VPG, while 85.3% were within EGA Zones A + B (clinically acceptable). MARD was 27.5% (95% CI: 20.8, 34.2%), with higher MARD values in the hypoglycaemia range and when VPG concentrations were falling. The FreeStyle Libre2TM CGM system tends to overestimate glucose concentrations compared to venous plasma samples in healthy women, especially during hypoglycaemia and during glycaemic swings.

Racial and Ethnic Differences in the Association Between Mean Glucose and Hemoglobin A1c.

Background: Studies have reported significantly higher hemoglobin A1c (A1C) in African American patients than in White patients with the same mean glucose, but less is known about other racial/ethnic groups. We evaluated racial/ethnic differences in the association between mean glucose, based on continuous glucose monitor (CGM) data, and A1C. Methods: Retrospective study among 1788 patients with diabetes from Kaiser Permanente Northern California (KPNC) who used CGM devices during 2016 to 2021. In this study population, there were 5264 A1C results; mean glucose was calculated from 124,388,901 CGM readings captured during the 90 days before each A1C result. Hierarchical mixed models were specified to estimate racial/ethnic differences in the association between mean glucose and A1C. Results: Mean A1C was 0.33 (95% confidence interval: 0.23-0.44; P < 0.0001) percentage points higher among African American patients relative to White patients for a given mean glucose. A1C results for Asians, Latinos, and multiethnic patients were not significantly different from those of White patients. The slope of the association between mean glucose and A1C did not differ significantly across racial/ethnic groups. Variance for the association between mean glucose and A1C was substantially greater within groups than between racial/ethnic groups (65% vs. 9%, respectively). Conclusions: For African American patients, A1C results may overestimate glycemia and could lead to premature diabetes diagnoses, overtreatment, or invalid assessments of health disparities. However, most of the variability in the mean glucose-A1C association was within racial/ethnic groups. Treatment decisions driven by guideline-based A1C targets should be individualized and supported by direct measurement of glycemia.

Mixed Meal Tolerance Test Versus Continuous Glucose Monitoring for an Effective Diagnosis of Persistent Post-Bariatric Hypoglycemia.

Gastric bypass determines an increase in incretin secretion and glucose excursions throughout the day and may sometimes entail the development of severe post-bariatric hypoglycemia (PBH). However, there is no consensus on the gold standard method for its diagnosis. In this study, we evaluated the usefulness of a mixed meal tolerance test (MMTT) and continuous glucose monitoring (CGM) for the diagnosis of PBH, defined as glucose levels <54 mg/dL (3.0 mmol/L). We found that hypoglycemia occurred in 60% of patients after the MMTT and in 75% during CGM, and it was predominantly asymptomatic. The MMTT confirmed the diagnosis of PBH in 88.9%of patients in whom surgery had been performed more than three years ago, in comparison to 36.4% in cases with a shorter postsurgical duration. CGM diagnosed nocturnal asymptomatic hypoglycemia in 70% of patients, and daytime postprandial hypoglycemia in 25% of cases. The mean duration of asymptomatic hypoglycemia was more than 30 min a day. Patients with ≥2% of their CGM readings with hypoglycemia exhibited a higher degree of glucose variability than those with <1% of the time in hypoglycemia. Our results show that the MMTT may be a useful dynamic test to confirm the occurrence of hypoglycemia in a large number of patients with persistent and recurrent PBH during long-term follow-up after gastric bypass. CGM, on its part, helps identify hypoglycemia in the real-world setting, especially nocturnal asymptomatic hypoglycemia, bringing to light that PBH is not always postprandial.

1816-PUB: Training and Support for Hospital Staff in the Use of Continuous Glucose Monitoring as Standard of Care in a Real-World Hospital Setting

Use of continuous glucose monitoring (CGM) in the hospital is limited and staff familiarity with CGM technology is low. During the COVID-19 pandemic and following the non-objection of the FDA to use of CGM in the hospital, the CGM as “standard of care” (CGM as SOC) program was implemented at Scripps Mercy Hospital in San Diego, CA using the Dexcom G6 CGM system. The implementation process of CGM as SOC included generating daily eligible patient lists, deploying specialized teams for CGM placement and connecting to nursing station iPads and centralized monitoring dashboards, and training of hospital nursing staff for CGM readings. A remote Advanced Practice Nurse (APN) served as an on-site trainer and expert consultant for troubleshooting. A Learning Management System (LMS) module was created for staff training and surveys assessed the training process. From May 2020-December 2022, CGM devices were placed on over 1200 adult hospitalized patients with diabetes requiring insulin. The APN conducted 30 live trainings for all units and shifts over 6 months. N=838 RNs completed the LMS module and met the required post-test passing rate of ≥ 80%. A total of N=38 self-report surveys were completed by nursing staff. Survey responses confirmed that most RNs felt confident in using the CGM nursing guide (97%) and had access to experts for troubleshooting (89%). Most common questions were related to technology troubleshooting (71%) compared to clinical issues (29%). Daily bedtime validation was transitioned from the APN to the bedside nursing staff using a validation calculator within EPIC. Despite limited baseline knowledge of CGM by hospital staff, deployment of specialized teams, APN on-site training and delivery of LMS modules supported the rapid implementation of CGM as SOC in a large community-based hospital. Standard training tools and processes can allow replication in health systems for use as part of the current provisional FDA allowance. Disclosure S.Lohnes: Consultant; Dexcom, Inc. R.Acuna: None. S.M.Saenz: None. T.Juarez: None. A.Philis-tsimikas: Advisory Panel; Dexcom, Inc., Novo Nordisk A/S, Sanofi, Other Relationship; Medtronic, Research Support; Novo Nordisk A/S, Lilly, Viking Therapeutics, NIH - National Institutes of Health. A.Bastian: None. A.L.Fortmann: Employee; Dexcom, Inc. L.Talavera: None. R.Belasco: None. H.Sandoval: None. M.Chichmarenko: None. S.R.Spierling bagsic: None. E.Soriano: None.

940-P: Real-World Concordance between Point-of-Care and Continuous Glucose Monitoring among Hospitalized Patients with Diabetes

Concordance between continuous glucose monitoring (CGM) and point-of-care testing (POCT) in research studies has been promising; however, the accuracy of using CGM for glucose management in real-world settings requires evaluation. Amid the COVID-19 pandemic, the CGM as “standard of care” (CGM as SOC) program was implemented at Scripps Mercy Hospital in San Diego, CA. In this real-world setting, concordance between the Dexcom G6 CGM system and POCT was evaluated among hospitalized adult patients with diabetes requiring insulin from May 2020-July 2022. For N=880 admissions, CGM readings over the first 24 hours of wear were analyzed relative to POCT conducted within 5 minutes. Patients had a median of 3 CGM-POCT data pairs, and the first CGM-POCT pair occurred a median of 2.75 hours after CGM device placement. Mean absolute relative difference (MARD) decreased over the first 24 hours of wear (B= -0.09%, p=0.0005; Figure 1A). Following the %20/20 rule, 62% of CGM readings were concordant upon the first measured POCT; 15.8% did not meet criteria at all during the first 24-hours. Using the Clarke Error Grid to describe concordance between CGM-POCT pairs (n=3015), only 0.7% (n=22) of CGM readings would have led to incorrect treatment decisions indicated by zones D/E (Figure 1B). These findings provide support for the accuracy of CGM for glucose management in the hospital. Disclosure S.R.Spierling bagsic: None. L.Talavera: None. A.Philis-tsimikas: Advisory Panel; Dexcom, Inc., Novo Nordisk A/S, Sanofi, Other Relationship; Medtronic, Research Support; Novo Nordisk A/S, Lilly, Viking Therapeutics, NIH - National Institutes of Health. A.L.Fortmann: Employee; Dexcom, Inc. R.Belasco: None. A.Ritko: None. S.Lohnes: Consultant; Dexcom, Inc. H.Sandoval: None. M.Chichmarenko: None. E.Soriano: None. A.Bastian: None.

958-P: Feasibility of a Long-Term Implanted Continuous Glucose Monitor (CGM)

Aim: To generate pilot data regarding acceptance, safety, and function of a novel long-term CGM. Methods: The investigational CGM (Eclipse 3, GlySens, San Diego, CA) was surgically inserted superficial to the rectus sheath in 8 type 1 diabetes adults (4 Female; Mean [SD] age 56 [7]; BMI 25 [3]). CGM data were masked to participants and processed retrospectively in simulated real-time with one capillary blood glucose calibration per day. Monthly visits evaluated: user acceptance by standardised questionnaire; health of implantation sites; safety by anti-glucose oxidase and anti-catalase antibodies; and performance benchmarked against YSI during standardised meal tests. Bluetooth communication between the sensor and an Apple iPhone was assessed continuously. Results: One sensor was removed at 4 months (participant withdrawal) due to impacted implant environment. There were no SAE’s. User acceptance remained high for the study duration. (Figure 1) Antibody levels did not rise. The lag corrected (23.3minutes) MARD (Median [range]) of 12.4% (1401 paired readings) was stable from month 1 (13.1%) to month 10 (14.0%). CGM readings within 20/20% and 30/30% of YSI values were 85% and 96%, respectively. The App received 95% of sensor data within 10 minutes. Conclusions: Our results provide initial data supporting participant acceptance, and safety of Eclipse 3. Further sensor refinement will reduce lag and improve performance. Disclosure A.Butler: Employee; GlySens Incorporated. J.Tan: None. D.N.O'neal: None. L.Robinson: None. V.R.Obeyesekere: None. H.Jones: None. Y.W.Kong: None. C.Yuan: None. K.Bertsch: Employee; GlySens Inc. T.L.Routh: Employee; GlySens Incorporated. S.L.Martha: Employee; GlySens Incorporated.

911-P: Changes in Metrics of Continuous Glucose Monitoring during COVID-19 in Children and Adolescents with Type 1 Diabetes

Background: Data on the changes in glucose control in pediatric patients with type 1 diabetes (T1D) affected by COVID-19 is lacking. We aimed to evaluate changes in metrics of continuous glucose monitoring (CGM) system during COVID-19 infection in children and adolescents with T1D. Methods: We included 21 children and adolescents with T1D, aged younger than 20 years, with at least 70% of CGM readings during COVID-19. Symptoms of COVID-19 infection were investigated through a questionnaire. The effects of COVID-19 on metrics of CGM were assessed at five different time points [2 weeks before-COVID-19 (Time 1), 1 week before-COVID-19 (Time 2), 1 week during COVID-19 (Time 3), 1 week after COVID-19 (Time 4), and 2 weeks after COVID-19 (Time 5)]. Results: Of 21 patients (6 males and 15 females), mean age was 11.4 ± 4.7 years, and mean disease duration was 5.1 ± 4.8 years. All had at least one related symptoms (mean duration of symptoms of 5.2 ± 3.7 days); no one required hospitalization. Glucose management indicator (GMI) and the percentage of time spent above range (&amp;gt; 250 mg/dL) were significantly higher at Time 3 compared to Time 5 (7.8 ± 1.4 % vs 7.2 ± 1.4 %; P = 0.001 and 18.3 ± 18.5 % vs 13.0 ± 15.7 %; P = 0.016, respectively). According to the insulin delivery methods, GMI at Time 2 was significantly higher compared to Time 5 in patients treated with multiple daily injection (MDI) (n = 13) but not in those using continuous subcutaneous insulin infusion (CSII) (n = 6). Conclusion: Pediatric patients with T1D experience mild COVID-19 infection with worsening glycemic control during COVID-19. The use of CSII is more effective in maintaining stable glycemic control during COVID-19 infection compared to MDI therapy. Disclosure H.Kim: None. S.Shin: None. S.Kim: None. J.Kim: None.

Fitterfly Diabetes CGM Digital Therapeutics Program for Glycemic Control and Weight Management in People With Type 2 Diabetes Mellitus: Real-world Effectiveness Evaluation.

Digital therapeutic platforms facilitate health care through patient-centered strategies based on multidisciplinary teams and shared decision-making. Such platforms can be used for developing a dynamic model of diabetes care delivery, which can help in improving glycemic control by promoting long-term behavior changes in people with diabetes. This study aims to evaluate the real-world effectiveness of the Fitterfly Diabetes CGM digital therapeutics program for improving glycemic control in people with type 2 diabetes mellitus (T2DM) after the completion of 90 days in the program. We analyzed deidentified data of 109 participants in the Fitterfly Diabetes CGM program. This program was delivered through the Fitterfly mobile app coupled with continuous glucose monitoring (CGM) technology. This program consists of 3 phases: the first phase is observation, wherein the patient's CGM readings are observed for 7 days (week 1); the second phase is the intervention; and the third phase aims at sustaining the lifestyle modification introduced during the second phase. The primary outcome of our study was the change in the participants' hemoglobin A1c (HbA1c) levels after program completion. We also evaluated the changes in participant weight and BMI after the program, changes in the CGM metrics in the initial 2 weeks of the program, and the effects of participant engagement in the program on improving their clinical outcomes. At the end of the 90 days of the program, the mean HbA1c levels, weight, and BMI of the participants were significantly reduced by 1.2% (SD 1.6%), 2.05 (SD 2.84) kg, and 0.74 (SD 1.02) kg/m2 from baseline values of 8.4% (SD 1.7%), 74.45 (SD 14.96) kg, and 27.44 (SD 4.69) kg/m2 in week 1, respectively (P<.001). The average blood glucose levels and time above range values showed a significant mean reduction by 16.44 (SD 32.05) mg/dL and 8.7% (SD 17.1%) in week 2 from week 1 baseline values of 152.90 (SD 51.63) mg/dL and 36.7% (SD 28.4%), respectively (P<.001 for both). Time in range values significantly improved by 7.1% (SD 16.7%) from a baseline value of 57.5% (SD 25%) in week 1 (P<.001). Of all the participants, 46.9% (50/109) showed HbA1c reduction ≥1% and 38.5% (42/109) showed weight loss ≥4%. The average number of times the mobile app was opened by each participant during the program was 108.80 (SD 127.91) times. Our study shows that participants in the Fitterfly Diabetes CGM program showed a significant improvement in their glycemic control and reduction in weight and BMI. They also showed a high level of engagement with the program. Weight reduction was significantly associated with higher participant engagement with the program. Thus, this digital therapeutic program can be considered as an effective tool for improving glycemic control in people with T2DM.

Continuous glucose monitoring in individuals undergoing gestational diabetes screening

Among guidelines on gestational diabetes mellitus, there is an incongruity about the threshold of maternal hyperglycemia to diagnose gestational diabetes mellitus. This study aimed to ascertain the association between continuous glucose monitoring metrics and adverse outcomes among individuals undergoing gestational diabetes mellitus screening. This was a prospective study (from June 2020 to January 2022) of individuals who underwent 2-step gestational diabetes mellitus screening at ≤30 weeks of gestation. The participants wore a blinded continuous glucose monitoring device (Dexcom G6 Pro; Dexcom, Inc, San Diego, CA) for 10 days starting when they took the 50-g glucose challenge test. The primary outcome was a composite of adverse neonatal outcomes (large for gestational age, shoulder dystocia or neonatal injury, respiratory distress, need for intravenous glucose treatment for hypoglycemia, or fetal or neonatal death). The secondary neonatal outcomes included preterm birth, neonatal intensive care unit admission, hypoglycemia, mechanical ventilation or continuous positive airway pressure, hyperbilirubinemia, and hospital length of stay. The secondary maternal outcomes included weight gain during pregnancy, hypertensive disorders of pregnancy, induction of labor, cesarean delivery, and postpartum complications. Time within the target range (63-140 mg/dL), time above the target range (>140 mg/dL) expressed as a percentage of all continuous glucose monitoring readings, and mean glucose level were analyzed. The Youden index was used to choose the threshold of ≥10% for the time above the target range and association with adverse outcomes. Of 136 participants recruited, data were available from 92 individuals (67.6%). The 2-step method diagnosed gestational diabetes mellitus in 2 individuals (2.2%). Continuous glucose monitoring indicated that 17 individuals (18.5%) had time above the target range of ≥10%. Individuals with time above the target range of ≥10% had a significantly higher likelihood of composite adverse neonatal outcomes than individuals with time above the target range of <10% (63% vs 18%; P=.001). Furthermore, compared with neonates born to individuals with time above the target range of <10%, neonates born to individuals with time above the target rangeof ≥10% had an increased likelihood for hypoglycemia (14.5% vs 47%; P=.009) and had a longer length of stay (2 vs 4 days; P=.03). No difference in maternal outcomes was noted between the groups. In this prospective study of individuals undergoing gestational diabetes mellitus screening, a cutoff of the time above the target range of ≥10% using continuous glucose monitoring was associated with a higher rate of neonatal adverse outcomes. A randomized trial of continuous glucose monitoring vs 2-step screening for gestational diabetes mellitus to lower the rate of adverse outcomes is underway (identification number: NCT05430204).

Hypoglycemia Detection Using Hand Tremors: Home Study of Patients With Type 1 Diabetes

Diabetes affects millions of people worldwide and is steadily increasing. A serious condition associated with diabetes is low glucose levels (hypoglycemia). Monitoring blood glucose is usually performed by invasive methods or intrusive devices, and these devices are currently not available to all patients with diabetes. Hand tremor is a significant symptom of hypoglycemia, as nerves and muscles are powered by blood sugar. However, to our knowledge, no validated tools or algorithms exist to monitor and detect hypoglycemic events via hand tremors. In this paper, we propose a noninvasive method to detect hypoglycemic events based on hand tremors using accelerometer data. We analyzed triaxial accelerometer data from a smart watch recorded from 33 patients with type 1 diabetes for 1 month. Time and frequency domain features were extracted from acceleration signals to explore different machine learning models to classify and differentiate between hypoglycemic and nonhypoglycemic states. The mean duration of the hypoglycemic state was 27.31 (SD 5.15) minutes per day for each patient. On average, patients had 1.06 (SD 0.77) hypoglycemic events per day. The ensemble learning model based on random forest, support vector machines, and k-nearest neighbors had the best performance, with a precision of 81.5% and a recall of 78.6%. The results were validated using continuous glucose monitor readings as ground truth. Our results indicate that the proposed approach can be a potential tool to detect hypoglycemia and can serve as a proactive, nonintrusive alert mechanism for hypoglycemic events.

Evaluation of a Fourth-Generation Subcutaneous Real-Time Continuous Glucose Monitor (CGM) in Individuals With Diabetes on Peritoneal Dialysis.

To evaluate the performance of a real-time continuous glucose monitor (CGM) in individuals with diabetes on peritoneal dialysis (PD). Thirty participants with type 2 diabetes on continuous ambulatory PD wore a Guardian Sensor 3 on the upper arm paired with Guardian Connect for 14 days. We compared CGM readings against Yellow Springs Instrument (YSI) venous glucose during an 8-h in-clinic session with glucose challenge. The mean absolute relative difference (MARD) was 10.4% (95% CI 9.6, 11.7) from 941 CGM-YSI matched pairs; 81.3% of readings were within %15/15 of YSI values in the full glycemic range. Consensus error grid analysis showed 99.9% of sensor values in zones A and B. There were no correlations between pH, uremia, hydration status, and MARD. We showed satisfactory performance of a real-time CGM sensor in PD patients with diabetes, supporting future use to facilitate treatment decisions.

Integration of multiomics with precision nutrition for gestational diabetes: Study protocol for the Westlake Precision Birth Cohort

Integration of multiomics with precision nutrition for gestational diabetes: Study protocol for the Westlake Precision Birth Cohort

Feasibility of intraoperative continuous glucose monitoring: An observational study in general surgery patients

BackgroundPerioperative hyperglycemia is associated with adverse outcomes in surgical patients, and major societies recommend intraoperative monitoring and treatment targeting glucose <180–200 mg/dL. However, compliance with these recommendations is poor, in part due to fear of unrecognized hypoglycemia. Continuous Glucose Monitors (CGMs) measure interstitial glucose with a subcutaneous electrode and can display the results on a receiver or smartphone. Historically CGMs have not been utilized for surgical patients. We investigated the use of CGM in the perioperative setting compared to current standard practices. MethodThis study evaluated the use of Abbott Freestyle Libre 2.0 and/or Dexcom G6 CGMs in a prospective cohort of 94 participants with diabetes mellitus undergoing surgery of ≥3 h duration. CGMs were placed preoperatively and compared to point of care (POC) BG checks obtained by capillary samples analyzed with a NOVA glucometer. Frequency of intraoperative blood glucose measurement was at the discretion of the anesthesia care team, with a recommendation of once per hour targeting BG of 140–180 mg/dL. Of those consented, 18 were excluded due to lost sensor data, surgery cancellation, or rescheduling to a satellite campus resulting in 76 enrolled subjects. There were zero occurrences of failure with sensor application. Paired POC BG and contemporaneous CGM readings were compared with Pearson product-moment correlation coefficients, and Bland-Altman plots. ResultsData for use of CGM in perioperative period was analyzed for 50 participants with Freestyle Libre 2.0, 20 participants with Dexcom G6, and 6 participants with both devices worn simultaneously. Lost sensor data occurred in 3 participants (15%) wearing Dexcom G6, 10 participants wearing Freestyle Libre 2.0 (20%) and 2 of the participants wearing both devices simultaneously. The overall agreement of the two CGM's utilized had a Pearson correlation coefficient of 0.731 in combined groups with 0.573 in Dexcom arm evaluating 84 matched pairs and 0.771 in Libre arm with 239 matched pairs. Modified Bland-Altman plot of the difference of CGM and POC BG indicated for the overall dataset a bias of −18.27 (SD 32.10). ConclusionsBoth Dexcom G6 and Freestyle Libre 2.0 CGMs were able to be utilized and functioned well if no sensor error occurred at time of initial warmup. CGM provided more glycemic data and further characterized glycemic trends more than individual BG readings. Required time of CGM warm up was a barrier for intraoperative use as well as unexplained sensor failure. CGMs had a fixed warm of time, 1 h for Libre 2.0 and 2 h for Dexcom G6 CGM, before glycemic data obtainable. Sensor application issues did not occur. It is anticipated that this technology could be used to improve glycemic control in the perioperative setting. Additional studies are needed to evaluate use intraoperatively and assess further if any interference from electrocautery or grounding devices may contribute to initial sensor failure. It may be beneficial in future studies to place CGM during preoperative clinic evaluation the week prior to surgery. Use of CGMs in these settings is feasible and warrants further evaluation of this technology on perioperative glycemic management.

The Official Journal of ATTD Advanced Technologies &amp; Treatments for Diabetes Conference 22‐25 February 2023 I Berlin &amp; Online

The Official Journal of ATTD Advanced Technologies &amp; Treatments for Diabetes Conference 22‐25 February 2023 I Berlin &amp; Online

Chinese diabetes datasets for data-driven machine learning

Data of the diabetes mellitus patients is essential in the study of diabetes management, especially when employing the data-driven machine learning methods into the management. To promote and facilitate the research in diabetes management, we have developed the ShanghaiT1DM and ShanghaiT2DM Datasets and made them publicly available for research purposes. This paper describes the datasets, which was acquired on Type 1 (n = 12) and Type 2 (n = 100) diabetic patients in Shanghai, China. The acquisition has been made in real-life conditions. The datasets contain the clinical characteristics, laboratory measurements and medications of the patients. Moreover, the continuous glucose monitoring readings with 3 to 14 days as a period together with the daily dietary information are also provided. The datasets can contribute to the development of data-driven algorithms/models and diabetes monitoring/managing technologies.

The association between continuous glucose metrics and adverse outcomes in individuals undergoing gestational diabetes screening

To ascertain the association between continuous glucose monitoring (CGM) metrics and adverse outcomes among individuals undergoing gestational diabetes (GDM) screening. A prospective study (06/2020 to 01/2022) of individuals who underwent 2-step GDM screening at ≤ 30 weeks. Participants had a blinded CGM placed (Dexcom G6PRO) for 10 days during their glucose challenge test. Primary outcome was a composite of adverse neonatal outcomes including large for gestational age, shoulder dystocia or neonatal injury, respiratory distress, hypoglycemia, or fetal and neonatal death. Secondary neonatal outcomes included NICU admission, need for intravenous glucose, mechanical ventilation or continuous positive airway pressure, hyperbilirubinemia, and hospital length of stay. We also evaluated maternal outcomes related to glucose control. Time in target range (TIR) (70–140 mg/dL), time above target range (TAR) ( >140mg/dL) expressed as percentage of all CGM readings, and mean glucose were analyzed. Youden index was used to choose cut point of ≥10% for TAR and association with outcomes. Of the 136 participants recruited, 45 (33%) did not return the CGM. The study focuses on the 91 (67%) with CGM data and among them 2-step method diagnosed GDM in 2 individuals (2.2%) (Table 1). CGM indicated that 16 (17.6%) had TAR≥10%. Individuals with TAR ≥10% had increased risk of composite adverse neonatal outcome (80% vs. 26%, p< 0.001) compared to those with TAR< 10% (Table 2). Furthermore, neonates born to individuals with TAR≥10% were born at an earlier gestational age, had increased IV glucose treatment (43% vs. 9%, p=0.004) and longer length of stay (4 vs. 2 days, p=0.007). No difference in maternal outcomes was found between the groups. In this prospective study of individuals undergoing GDM screening a cutoff of TAR of ≥10% using CGM was associated with a higher rate of neonatal adverse outcomes. A randomized trial of CGM vs. 2-step screening for GDM to lower the rate of adverse outcomes is planned (NCT05430204).View Large Image Figure ViewerDownload Hi-res image Download (PPT)