bims-glumda Biomed News
on CGM data in management of diabetes
Issue of 2026–03–08
ten papers selected by
Mott Given
  1. Evaluation of a New Continuous Glucose Monitoring Device.
  2. Future-aware blood glucose forecasting using knowledge distillation with transformer-based sequence-to-sequence models.
  3. Photocrosslinked dextran hydrogel coatings with FAD-glucose dehydrogenase for continuous glucose monitoring in complex media: Key role of methacrylation degree on bioelectrocatalytic activity and sensor stability.
  4. Modeling strategies for CGM data: A scoping review of mechanistic, machine learning, and hybrid approaches in diabetes management.
  5. Discordance Between HbA1c and Glucose Management Indicator in People With Type 2 Diabetes Treated With SGLT2 Inhibitors.
  6. Glucose dysregulation in hospitalized non-critically ill patients with a suspected infection: A prospective study using continuous glucose monitoring.
  7. Long-Term Growth Outcomes of Children With Type 1 Diabetes According to Glycemic Control and Use of Continuous Glucose Monitoring: A Retrospective Cohort Study.
  8. Impact of Continuous Glucose Monitor on Health Care Resource Utilization in Adults with Type 2 Diabetes Managed by Noninsulin Therapies.
  9. Clinical characteristics and associated factors of constipation patients with type 2 diabetes.
  10. Temporal gradient analysis of blood glucose responses to non-standard physical activity: a free-living study in type 1 diabetes.
  1. J Diabetes Sci Technol. 2026 Mar 06. 19322968261427025
    Evaluation of a New Continuous Glucose Monitoring Device.
    Matthew Kenneth Lancaster, Hope Edwards, Dawn Harper, Pip Garner, Andrea Utley.
      
    Keywords:  MARD; accuracy; bluetooth; continuous glucose monitor; diabetes
    DOI:  https://doi.org/10.1177/19322968261427025
  2. Sci Rep. 2026 Feb 28.
    Future-aware blood glucose forecasting using knowledge distillation with transformer-based sequence-to-sequence models.
    Xiaoyu Sun, Hongru Li, Xia Yu.
      Accurate blood glucose level (BGL) forecasting is critical for diabetes self-management and clinical decision-making. Although deep learning models based on continuous glucose monitoring (CGM) data have achieved encouraging results, most approaches rely exclusively on historical observations and cannot explicitly account for future disturbances, such as insulin delivery and meal intake, that are unavailable at deployment. To address this limitation, we propose a future-aware learning framework for multi-step BGL prediction that leverages privileged information during training while preserving deployability at inference. A Transformer-based teacher model is trained offline using both historical CGM data and future disturbance information to learn disturbance-aware temporal representations. A student model with a similar sequence-to-sequence structure is then trained using knowledge distillation to approximate the teacher's representations based solely on historical inputs, enabling real-time forecasting without access to future data. The proposed framework is evaluated on the publicly available OhioT1DM and AZT1D datasets for prediction horizons ranging from 30 to 120 minutes and compared with several established methods. The results show consistent reductions in root mean squared error and mean absolute error, together with improved clinical reliability as assessed by Clarke error grid analysis, with over 90% of predictions falling within clinically acceptable regions. These findings demonstrate the potential of future-aware training strategies to enhance glucose forecasting performance under realistic deployment constraints.
    DOI:  https://doi.org/10.1038/s41598-026-41787-7
  3. Biosens Bioelectron. 2026 Feb 28. pii: S0956-5663(26)00207-1. [Epub ahead of print]303 118575
    Photocrosslinked dextran hydrogel coatings with FAD-glucose dehydrogenase for continuous glucose monitoring in complex media: Key role of methacrylation degree on bioelectrocatalytic activity and sensor stability.
    Bastien Darmau, Vladyslav Mishyn, Abir Elloumi, Isabelle Texier, Andrew J Gross.
      Fungal flavin-dependent glucose dehydrogenase (FAD-GDH) is now widely preferred as an O2-insensitive alternative to glucose oxidase for 2nd generation blood glucose test strips. FAD-GDH bioelectrodes have potential for continuous glucose monitoring (CGM), but continue to be hampered by poor operational stability, restricted mediator compatibility, and selectivity limitations. Herein, we report new protective biosensor coatings based on covalently photocrosslinked polysaccharides for more robust CGM with FAD-GDH bioelectrodes. The crosslinked hydrogel membranes were prepared from dextran methacrylate (Dex-MA) polymers synthesised with different degrees of substitution (DS = 9%, 18%, and 37%). The polymers were dip-coated then crosslinked via a photoinitiator using a rapid visible light process (λ = 405 nm; 1 min). This study highlights the crucial impact of the polymer DS on redox mediator electroactivity, O2 reactivity, catalytic glucose activity, storage stability and operational stability. A higher polymer DS provided improved mediator stabilisation and up to a 4-fold increase in 1-week storage stability. A high DS of 37% also significantly increased CGM stability and permitted attractive sensor analytics in artificial interstitial fluid (ISF). The three sensors prepared with a DS of 9% to 37% provided practical linear ranges and detection limits for CGM. A CGM lifetime of 54 h was achieved in a complex artificial ISF comprising electroactive interferences, compared to only 16 h for an equivalent biosensor without hydrogel protection. Photocrosslinked polysaccharide hydrogel membranes hold promise for extending bioelectrocatalytic outputs for future biosensors and eventually biofuel cells and bioreactors.
    Keywords:  Biopolymers; Continuous glucose monitoring; Diabetes; Electrochemical sensors; Wearable biosensors
    DOI:  https://doi.org/10.1016/j.bios.2026.118575
  4. Comput Biol Med. 2026 Mar 03. pii: S0010-4825(26)00164-2. [Epub ahead of print]205 111601
    Modeling strategies for CGM data: A scoping review of mechanistic, machine learning, and hybrid approaches in diabetes management.
    Nilde Fera, Anna Procopio, Paolo Zaffino, Antonio Cutruzzolà, Concetta Irace, Carlo Cosentino.
       BACKGROUND: The increasing availability of continuous glucose monitoring (CGM) data has opened new avenues for modeling glucose dynamics in diabetes management.
    OBJECTIVE: This scoping literature review uniquely explores the full methodological spectrum applied to CGM data analysis, ranging from classical Mechanistic Models (MM) and statistical time series approaches, to modern Artificial Intelligence (AI) techniques, and emerging hybrid frameworks that combine the two paradigms. Unlike prior reviews focused solely on Type 1 Diabetes Mellitus (T1DM), the present work includes modeling efforts across different classes of populations, including Type 2 Diabetes Mellitus (T2DM), Gestational Diabetes Mellitus (GDM), and other forms of diabetes.
    METHODS: Literature was systematically retrieved from Elsevier Scopus®, Clarivate Web of Science™, and PubMed®, providing a comprehensive and comparative assessment of state-of-the-art strategies for CGM-based analysis in diverse clinical contexts.
    RESULTS: The reviewed studies indicate a clear methodological shift toward data-driven and hybrid frameworks. Overall, both mechanistic and AI-based models achieve satisfactory predictive performance; however, their strengths differ across tasks. Machine learning methods are particularly effective for event detection and feature extraction, whereas deep learning models excel in forecasting CGM-derived glucose trajectories. Hybrid approaches further enhance predictive accuracy while preserving physiological interpretability, especially over longer prediction horizons.
    CONCLUSION: Challenges such as data heterogeneity, the limited availability of high-quality datasets beyond T1DM, and reduced cross-cohort generalizability persist, underscoring the need for standardized validation procedures and physiologically informed modeling strategies.
    Keywords:  Continuous glucose monitoring; Diabetes; Physiological models
    DOI:  https://doi.org/10.1016/j.compbiomed.2026.111601
  5. J Diabetes Sci Technol. 2026 Feb 28. 19322968261426379
    Discordance Between HbA1c and Glucose Management Indicator in People With Type 2 Diabetes Treated With SGLT2 Inhibitors.
    Akinori Hayashi, Tomoaki Tsukushi, Ryota Shinohara, Haruna Sogabe, Satoshi Tsuji, Naoya Shimizu, Takeshi Miyatsuka.
       BACKGROUND: The hemoglobin glycation index (HGI), defined as the difference between HbA1c and the glucose management indicator (GMI) derived from continuous glucose monitoring (CGM), has emerged as a tool to evaluate discordance between laboratory and sensor-based measures. The impact of sodium-glucose cotransporter 2 inhibitors (SGLT2i) on these markers remains unclear.
    METHODS: We retrospectively analyzed CGM data from 143 individuals with type 2 diabetes, stratified by SGLT2i use. Both HGI and glycated albumin-to-HbA1c (GA/HbA1c) ratio were compared. A restricted dataset (n = 117) excluding individuals with anemia or advanced renal dysfunction was also examined.
    RESULTS: SGLT2i users exhibited higher hematologic parameters and significantly greater HGI (full dataset: 0.3 vs 0.1, P = .0297; restricted: 0.4 vs 0.1, P = .0206), whereas the GA/HbA1c ratio was significantly lower (full: 2.27 vs 2.54, P = .0117; restricted: 2.18 vs 2.38, P = .0178). The HbA1c-GMI relationship showed significantly different slopes between SGLT2i users and non-users, while the GA-GMI relationship was consistent across groups. Multivariate analyses identified SGLT2i use as an independent determinant of both higher HGI and lower GA/HbA1c ratio, even after adjustment for age, sex, body mass index (BMI), hemoglobin, albumin, renal function, mean glucose, and glycemic variability.
    CONCLUSION: SGLT2i therapy alters the interpretation of glycemic markers by elevating HGI and lowering GA/HbA1c, independent of hematologic and renal factors. These findings emphasize the need for individualized assessment of glycemic control using CGM-derived metrics and complementary biomarkers.
    Keywords:  continuous glucose monitoring; glucose management indicator; hemoglobin A1c; hemoglobin glycation index; sodium–glucose cotransporter 2 inhibitors
    DOI:  https://doi.org/10.1177/19322968261426379
  6. PLoS One. 2026 ;21(3): e0343703
    Glucose dysregulation in hospitalized non-critically ill patients with a suspected infection: A prospective study using continuous glucose monitoring.
    Anna D Schoonhoven, Julia J Bakker, Alessandra D Di Mauro, Amarens van der Vaart, Riemer Been, André P van Beek, Pratik Choudhary, Hjalmar R Bouma, Peter R van Dijk.
       INTRODUCTION: Dysglycaemia, defined as hypo- or hyperglycaemia, can occur during infection and is associated with worse outcomes during hospitalization. Previous studies on dysglycaemia in non-critically ill patients on general wards used point-of-care (POC) capillary measurements, possibly underestimating the problem. We assessed the prevalence and course of dysglycaemia in this population using continuous glucose monitoring (CGM).
    METHODS: In this prospective, observational study at the University Medical Center Groningen, adults admitted to the Emergency Department with suspected infections were enrolled via the Acutelines data and biobank. Participants wore blinded CGM sensors (FreeStyle Libre Pro iQ) while continuing usual care. Episodes of dysglycaemia were defined as ≥15 minutes of glucose <3.9 mmol/L or >10 mmol/L. Primary outcome was the number of dysglycaemic episodes; secondary outcomes included duration, glucose levels, and associations with clinical outcomes.
    RESULTS: CGM data from 90 participants (27% with a history of diabetes and 73% without) over a median of 3.4 days revealed 181 hyperglycaemia and 303 hypoglycaemia episodes. In patients with a history of diabetes, 75% experienced hyperglycaemia (median of 6.5 events/patient). In contrast, 33% of individuals without prior diabetes experienced hyperglycaemia (median 1.5 events/ patient). Median Time in Range (glucose 3.9-10.0 mmol/L) was 59% for patients with and 86% for patients without known diabetes. Exploratory analyses showed no significant association between dysglycaemia and ICU admission or 30-day mortality.
    CONCLUSIONS: This observational study provides relevant insight into dysglycaemia among non-critically ill patients admitted to the hospital. Significant hyperglycaemia was observed in both participants with and without known diabetes. Therefore, CGM may enable earlier detection of dysglycaemia and thereby inform future interventional research and in-hospital strategies.
    DOI:  https://doi.org/10.1371/journal.pone.0343703
  7. Pediatr Diabetes. 2026 ;2026 9111583
    Long-Term Growth Outcomes of Children With Type 1 Diabetes According to Glycemic Control and Use of Continuous Glucose Monitoring: A Retrospective Cohort Study.
    Sujin Kim, Seo Jung Kim, Kyoung Won Cho, Kyungchul Song, Youngha Choi, Hyun Wook Chae, Junghwan Suh.
       Background: Chronic diseases such as type 1 diabetes mellitus (T1DM) may alter linear growth; however, reports regarding growth in children with T1DM have been inconsistent. This study aimed to investigate the height and growth velocity of patients with T1DM, and whether they were affected by various factors 5 years after the diagnosis.
    Methods: This retrospective study included patients with T1DM between October 2005 and May 2022, with a follow-up period of at least 1 year. Patients with diabetes, thyroid disease, celiac disease, or any other chronic disease were excluded. We compared the mean height standard deviation score (H-SDS) and growth velocity between groups divided based on glycosylated hemoglobin (HbA1c) levels and use of continuous glucose monitoring (CGM) systems.
    Results: Among the 150 patients, 45.3% were male, with a mean age at diagnosis of 7.8 ± 3.6 years. At diagnosis, the mean H-SDS was 0.38 ± 1.11. In males, H-SDS significantly decreased overtime, with an estimated slope (β) of -0.054 (standard error [SE] = 0.013, 95% confidence interval [CI]: -0.079 to -0.029, p  < 0.01). The decline in H-SDS was more pronounced in the poorly-controlled group (mean HbA1c ≥7.0%) compared to the well-controlled group (mean HbA1c <7.0%; β = -0.081, SE = 0.016, 95% CI: -0.112 to -0.050 vs. β = -0.007, SE = 0.020, 95% CI: -0.047 to -0.033, p  < 0.01). Among males using CGM, the decrease in H-SDS over the 5-year follow-up was significantly less than that observed in the non-CGM group (β = -0.012, SE = 0.023, 95% CI: -0.057 to -0.034 vs. β = -0.072, SE = 0.015, 95% CI: -0.101 to -0.042, p = 0.03). In the multivariable linear mixed model analysis, younger age at diagnosis (β = -0.009, 95% CI: -0.017 to -0.002, p = 0.02), female (β = 0.067, 95% CI: 0.033 to 0.100, p  < 0.01) and lower HbA1c levels (β = -0.026, 95% CI: -0.038 to -0.015, p < 0.01) were significantly associated with greater improvement in H-SDS over 5 years.
    Conclusion: Glycemic control and CGM use positively affected linear growth in children with T1DM, especially in males. CGM use was associated with improved growth outcomes, which suggests that glucose monitoring may help mitigate the adverse effects of poor glycemic control on growth.
    Keywords:  children; continuous glucose monitoring; glycemic control; height standard deviation score; linear growth; type 1 diabetes mellitus
    DOI:  https://doi.org/10.1155/pedi/9111583
  8. Diabetes Technol Ther. 2026 Mar 01. 15209156261423934
    Impact of Continuous Glucose Monitor on Health Care Resource Utilization in Adults with Type 2 Diabetes Managed by Noninsulin Therapies.
    Anup K Sharma, Abraham Lee, Shivani Mehta, Qaashif Panjwani, Bhavya Sree Burugapalli.
       BACKGROUND: Daily use of continuous glucose monitoring (CGM) has been shown to reduce diabetes-related events and associated costs in individuals with type 2 diabetes (T2D) regardless of their therapy. However, adoption of CGM among the large majority of T2D adults in the United States who are treated with noninsulin therapies has been limited.
    METHODS: This retrospective database study assessed the effects of CGM acquisition on health care resource utilization (HCRU) in a large cohort of T2D adults treated with noninsulin, antidiabetes therapies. Inclusion criteria were T2D diagnosis, age ≥18 years, treated with noninsulin therapies, CGM-naive before CGM acquisition, and continuous medical/pharmacy insurance coverage during the 12-month preindex and postindex periods. The primary outcome measures were changes in all-cause hospitalizations (ACH), emergency department (ED) visits, acute diabetes complications, hyperglycemic events (HGE), and diabetic ketoacidosis (DKA) during the 12 months following CGM acquisition.
    RESULTS: A total of 20,468 adults with T2D were included in this analysis. CGM acquisition was associated with significant reductions in event rates in the postindex period compared with the preindex period HCRU at 12 months: ACH (-25%), ED visits (-7%), HGE (-7%), DKA (-86%), and acute diabetes complications (-7%), all P < 0.0001. Similar reductions in events per person were also observed: ACH (-20.6%), ED visits (-7.2%), HGE (-6.2%), DKA (-63.0%), and acute diabetes complications (-6%), all P < 0.0001. Significant reductions were also seen in patients with cardiovascular disease. ACH and ED visits decreased by 35% and 12%, respectively; in those with liver disease by 25% and 13%; in those with renal disease, ACH decreased by 33%; and in those with hypertension, ACH and ED visits decreased by 26% and 7%, respectively. All reductions were statistically significant (P < 0.01).
    CONCLUSIONS: This analysis demonstrated an association between CGM acquisition and reductions in HCRU in adults with T2D treated with noninsulin therapies.
    Keywords:  diabetic ketoacidosis; emergency department; health care resource utilization; hospitalization; type 2 diabetes
    DOI:  https://doi.org/10.1177/15209156261423934
  9. Front Endocrinol (Lausanne). 2026 ;17 1698701
    Clinical characteristics and associated factors of constipation patients with type 2 diabetes.
    Yan Fang, Jiahao Chen, Chendong Ni, Yi Dai, Lili Cai, Xiaohua Yang, Ji Hu, Hong-Hong Zhang.
       Aim: The aim of this study was to explore the clinical characteristics and associated factors of constipation in patients with type 2 diabetes (T2DM). Here, we focus on the correlation between time in range (TIR) and coefficient of variation (CV) of glucose and constipation in patients with T2DM.
    Methods: In the exploratory, cross-sectional study, a total of 120 patients diagnosed with T2DM in the department of Endocrinology and Gastroenterology of Liqun hospital from 2023 to 2024 were recruited. Patients with concurrent constipation were included in the constipation group, and those without concurrent constipation were included in the non-constipation group. Fasting blood indicators of patients were detected, including fasting blood glucose (FBG), 2 hours postgrandial blood glucose (2hPBG), glycosylated hemoglobin type A1C (HbA1c), fasting C-peptide (FCP), homeostasismodel assessment-insulin resistance (HOMA-IR), triglyceride (TG), total cholesterol (TC), high density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C). Blood Glucose was monitored by a silicon-based dynamic system for 14 days, and TIR and CV were calculated by a continuous glucose monitoring system (CGM). In addition, the symptoms of constipation, gastroparesis cardinal symptom index (GCSI), patient assessment of upper gastrointestinal symptom severity index (PAGI-SYM) and patient assessment of quality-of-life index (PAGI-QOL) were evaluated by questionnaires. The impact of TIR and CV for constipation risk was evaluated using receiver operating characteristic (ROC) curves.
    Results: 1. More serious condition in constipation symptoms, GCSI, VSI, PAGI-SYM and PAGI-QOL scores in the constipation group. 2. Compared with the non-constipation group, the constipation group had longer courses, and higher probabilites of complications of autonomic neuropathy, diabetic nephropathy and hyperlipidemia; 3. Compared with the non-constipated group, FCP was significantly decreased, and HbA1c, TC and LDL-C were significantly increased in the constipation group. 4. TIR and CV were significantly correlated with constipation in patients with T2DM.
    Conclusions: Lower TIR and higher CV were associated with constipation in T2DM patients. These findings suggest that CGM-derived metrics may be useful markers of constipation risk and warrant investigation in prospective studies to determine whether improving these parameters can prevent constipation development.
    Keywords:  CGM; T2DM; coefficient of variation; constipation; glucose target time in range
    DOI:  https://doi.org/10.3389/fendo.2026.1698701
  10. Front Sports Act Living. 2026 ;8 1718510
    Temporal gradient analysis of blood glucose responses to non-standard physical activity: a free-living study in type 1 diabetes.
    Ahmad Bilal, Hood Thabit, Paul W Nutter, Simon Harper.
       Introduction: Daily physical activity (PA) impacts blood glucose (BG) in individuals with Type 1 Diabetes Mellitus (T1DM), with effects varying by intensity, duration, and timing. Predicting BG changes during free-living activity remains challenging but may help prevent hypoglycaemia. Previous studies have focused on the impact of PA on BG levels, but only during exercise sessions, not throughout the entire day.
    Methods: Using retrospective data from eight individuals with T1DM (mean age 67 years; 3 female, 5 male), we analysed whether non-standard PA, defined as activity exceeding the individual's mean habitual level in a preceding interval, was associated with steeper downward trends in BG. PA was quantified using wrist-worn accelerometry, and BG responses were analysed using gradient-based methods across 20, 40, and 60 min time windows.
    Results: Two hypotheses were evaluated. Hypothesis 1 assessed whether BG decline intensified during existing downward trends and achieved an accuracy above 83.33%, with F1-scores exceeding 0.83 at shorter intervals. Hypothesis 2 examined BG declines following prior increases and showed greater variability; accuracy ranged from 73.53% to 88.33%, with the lowest F1-score of 0.75 at the 60 min window.
    Conclusion: We have found a reliable correlation between increased levels of PA and BG levels under free-living conditions. These findings establish a foundation for future work aimed at quantifying BG responses to PA and developing personalised decision-support tools for insulin or carbohydrate adjustment.
    Keywords:  blood glucose; continuous glucose monitoring; free-living; gradient analysis; personalised modelling; physical activity; type 1 diabetes; wearable sensors
    DOI:  https://doi.org/10.3389/fspor.2026.1718510
This page is being maintained by Thomas Krichel. It was last updated on 2026‒03‒08 at 0:07.