Multi-Horizon Glucose Prediction Across Populations with Deep Domain Generalization.

Real-time continuous glucose monitoring (CGM), augmented with accurate glucose prediction, offers an effective strategy for maintaining blood glucose levels within a therapeutically appropriate range. This is particularly crucial for individuals with type 1 diabetes (T1D) who require long-term self-management. However, with extensive glycemic variability, developing a prediction algorithm applicable across diverse populations remains a significant challenge. Leveraging meta-learning for domain generalization, we propose GPFormer, a Transformer-based zero-shot learning method designed for multi-horizon glucose prediction. We developed GPFormer on the REPLACE-BG dataset, comprising 226 participants with T1D, and proceeded to evaluate its performance using three external clinical datasets with CGM data. These included the OhioT1DM dataset, a publicly available dataset including 12 T1D participants, as well as two proprietary datasets. The first proprietary dataset included 22 participants, while the second contained 45 participants, encompassing a diverse group with T1D, type 2 diabetes, and those without diabetes, including patients admitted to hospitals. These four datasets include both outpatient and inpatient settings, various intervention strategies, and demographic variability, which effectively reflect real-world scenarios of CGM usage. When compared with a group of machine learning baseline methods, GPFormer consistently demonstrated superior performance and achieved the lowest root mean square error for all the evaluated datasets up to a prediction horizon of two hours. These experimental results highlight the effectiveness and generalizability of the proposed model across a variety of populations, demonstrating its substantial potential to enhance glucose management in a wide range of practical clinical settings.

Identifying the deficiencies of currently available CGM to improve uptake and benefit.

BACKGROUND AND AIMS: The use of diabetes technologies is increasing worldwide, with health systems facilitating improved access to devices. Continuous glucose monitoring is a complex intervention that provides information on glucose concentration, rate and direction of change, historical data and alerts and alarms for extremes of glucose. These data do not themselves change glycaemia and require translation to a meaningful action for impact. It is, therefore, crucial that such systems advance to better meet the needs of individuals using them. METHODS: Narrative review of the use of, engagement with, limitations and unmet needs of continuous glucose monitoring systems. RESULTS: CGM devices have made a significant contribution to the self-management of diabetes; however, challenges with access and user experience persist, with multiple limitations to uptake and benefit. These limitations include physical size and implementation, with associated stigma, alarm fatigue, sleep disturbance and the challenge of addressing large volumes of real-time data. Greater personalisation throughout the continuous glucose monitoring journey, with a focus on usability, may improve the benefits derived from the device and reduce the burden of self-management. Healthcare professionals may have unconscious biases that affect the provision of continuous glucose monitors due to deprivation, education, age, ethnicity and other characteristics. CONCLUSIONS: Continuous glucose monitoring exerts a dose-dependent response; the more it is used, the more effective it is. For optimal use, continuous glucose monitors must not just reduce the burden of management in one dimension but facilitate net improvement in all domains of self-management for all users.

What Is the Influence of HIV Serostatus on HbA1c? A Prospective Analysis Using Continuous Glucose Monitoring.

OBJECTIVE: Type 2 diabetes (T2D) is reported to be more common in people living with HIV (PLWH). Clinical guidelines recommend screening for diabetes in PLWH, but there is no agreed method due to studies reporting HbA1c is falsely low in PLWH. These studies were performed in the early HIV era when participants were taking older preparations of antiretroviral therapy that are rarely used today. We aimed to investigate whether HIV serostatus influences HbA1c. RESEARCH DESIGNS AND METHODS: We conducted a prospective cohort study of PLWH and age- and sex-matched HIV-negative participants who were purposely recruited from clinics in Brighton, U.K. Each participant wore a Dexcom G6 continuous glucose monitor (CGM) for up to 10 days, had glucose measured during an oral glucose tolerance test, and fructosamine and paired HbA1c were measured. We performed regression analysis to assess the influence of HIV on HbA1c and used a separate model for CGM glucose, venous glucose, and fructosamine. In addition, we included predictor variables used in previous studies that explored HbA1c discrepancy. RESULTS: We recruited 60 PLWH (90% men, 50% with T2D, mean ± SD age 57 ± 10.7 years, 100% undetectable viral load) and 48 people without HIV (92% men, 30% with T2D, mean age 57.7 ± 8.9 years). We found that HIV serostatus did not have a significant influence on HbA1c within the regression models. CONCLUSIONS: We performed a comprehensive assessment of glycemia to assess whether HIV serostatus influences HbA1c. We did not find any strong evidence that HIV serostatus influenced HbA1c. The results of our study support incorporating HbA1c into routine clinical blood work in PLWH.

Psychometric validation of the hyperglycaemia avoidance scale UK (HAS-UK).

AIMS: Hyperglycaemia aversion in type 1 diabetes can be associated with severe hypoglycaemia and impaired awareness of hypoglycaemia but is not routinely assessed clinically. This study aimed to undertake the first psychometric validation of the UK version of the Hyperglycaemia Avoidance Scale (HAS-UK). METHODS: The HAS-UK was completed by adults with type 1 diabetes in three separate research studies. Psychometric properties were evaluated, using exploratory factor analysis, internal consistency, and convergent validity. RESULTS: Of the 431 participants who completed the HAS-UK in the three studies, mean age was 49.5 years, and 58.0% were women. Mean duration of diabetes was 29 years, with 192 (44.5%) using multiple daily injections and 229 (53.1%) using an insulin pump. Five participants were excluded from analyses due to incomplete HAS-UK responses. Exploratory factor analysis revealed a 3-factor solution, with acceptable internal consistency for 'worry' and 'blood glucose decisions' factors. HAS-UK total score was higher in those using insulin pumps versus multiple daily injections, and 'blood glucose decisions' score was higher in those using a continuous blood glucose sensor versus a meter. CONCLUSIONS: The HAS-UK is a reliable measure with acceptable structural validity and is likely to be useful for evaluating hyperglycaemia aversion in people with type 1 diabetes. Future research would benefit from investigating further psychometric properties including test-retest reliability, sensitivity to change, and clinical significance of scores.

A Three-Arm Randomized Controlled Study Comparing Patient-Reported Outcomes in People With Type 1 Diabetes Using Continuous Subcutaneous Insulin Infusion or Multiple Daily Injections.

BACKGROUND: The aim of this study was to compare patient-reported outcomes (PROs) in people with type 1 diabetes using either continuous subcutaneous insulin infusion (CSII) with two different insulin patch pumps or multiple daily injections (MDIs). MATERIALS AND METHODS: In this randomized three-arm study, people with type 1 diabetes on MDI therapy were included and used either MDI, the Accu-Chek Solo micropump system (Solo) or Omnipod for 26 weeks. From weeks 26 to 39, all participants used CSII with Solo. Patient-reported outcomes were assessed using the diabetes technology questionnaire (DTQ); in addition, HbA1c values were measured. RESULTS: Overall, 181 participants were randomized (61 MDI arm, 62 Solo arm, 58 Omnipod arm) and 142 completed the study. After 26 weeks in the study, the DTQ "change" score in the Solo group (105.9 [100.6-111.2]; baseline-adjusted mean [95% confidence interval]) was significantly higher than in the MDI group (94.8 [89.6-100.0]) (P = .001). The comparison between the Solo group (105.1 [99.1-111.1]) and the Omnipod group (108.7 [103.1-114.4]) showed no significant differences (P = .382). HbA1c increased by 0.2% ± 0.7% in the MDI group and decreased in both pump groups (Solo group -0.2% ± 0.8% and Omnipod group -0.1% ± 0.8%). Differences in HbA1c between the Solo group and the MDI group were significant (P = .009), but not between the Solo group and the Omnipod group (P = .896). CONCLUSIONS: This study showed that switching from MDI to CSII improves both psychosocial well-being and physiological outcomes. Furthermore, there were no substantial differences between the established and the recently released patch pump. Trial registration at www.clinicaltrials.gov is NCT03478969.

Assessment of the impact of a personalised nutrition intervention in impaired glucose regulation over 26 weeks: a randomised controlled trial.

Dietary interventions can reduce progression to type 2 diabetes mellitus (T2DM) in people with non-diabetic hyperglycaemia. In this study we aimed to determine the impact of a DNA-personalised nutrition intervention in people with non-diabetic hyperglycaemia over 26 weeks. ASPIRE-DNA was a pilot study. Participants were randomised into three arms to receive either (i) Control arm: standard care (NICE guidelines) (n = 51), (ii) Intervention arm: DNA-personalised dietary advice (n = 50), or (iii) Exploratory arm: DNA-personalised dietary advice via a self-guided app and wearable device (n = 46). The primary outcome was the difference in fasting plasma glucose (FPG) between the Control and Intervention arms after 6 weeks. 180 people were recruited, of whom 148 people were randomised, mean age of 59 years (SD = 11), 69% of whom were female. There was no significant difference in the FPG change between the Control and Intervention arms at 6 weeks (- 0.13 mmol/L (95% CI [- 0.37, 0.11]), p = 0.29), however, we found that a DNA-personalised dietary intervention led to a significant reduction of FPG at 26 weeks in the Intervention arm when compared to standard care (- 0.019 (SD = 0.008), p = 0.01), as did the Exploratory arm (- 0.021 (SD = 0.008), p = 0.006). HbA1c at 26 weeks was significantly reduced in the Intervention arm when compared to standard care (- 0.038 (SD = 0.018), p = 0.04). There was some evidence suggesting prevention of progression to T2DM across the groups that received a DNA-based intervention (p = 0.06). Personalisation of dietary advice based on DNA did not result in glucose changes within the first 6 weeks but was associated with significant reduction of FPG and HbA1c at 26 weeks when compared to standard care. The DNA-based diet was effective regardless of intervention type, though results should be interpreted with caution due to the low sample size. These findings suggest that DNA-based dietary guidance is an effective intervention compared to standard care, but there is still a minimum timeframe of adherence to the intervention before changes in clinical outcomes become apparent.Trial Registration: www.clinicaltrials.gov.uk Ref: NCT03702465.

The role of real-time continuous glucose monitoring in diabetes management and how it should link to integrated personalized diabetes management.

Diabetes is a complex metabolic condition that demands tailored, individualized approaches for effective management. Real-time continuous glucose monitoring (rtCGM) systems have improved in terms of design, usability and accuracy over the years and play a pivotal role in the delivery of integrated personalized diabetes management (iPDM). iPDM is a comprehensive multidisciplinary approach that combines individualized care strategies utilizing technologies and interventions and encourages the active involvement of the person with diabetes in the care provided. The use of stand-alone rtCGM and its integration with other diabetes technologies, such as hybrid automated insulin delivery, have enabled improved glycaemic and quality of life outcomes for people with diabetes. As the uptake of rtCGM and associated technologies is increasing and becoming the standard of care for people with diabetes, it is important that efforts are focused on associated goals such as reducing health inequalities in terms of access, aligning structured education with rtCGM usage, choosing the right technology based on needs and preferences, and minimizing burden while aiming for optimal glucose outcomes. Utilizing rtCGM in other settings than outpatients and in diabetes cohorts beyond type 1 and type 2 diabetes needs further exploration. This review aims to provide an overview of the role of rtCGM and how best to link rtCGM to iPDM, highlighting its role in enhancing personalized treatment strategies.

Ward monitoring 4.0: real-time metabolic insights from continuous glucose monitoring into perioperative organ dysfunction.

The now-routine clinical deployment of continuous glucose monitoring has demonstrated benefit in real-world settings. We make the case that continuous glucose monitoring can help re-examine, at scale, the role that (stress) hyperglycaemia plays in fuelling organ dysfunction after tissue trauma. Provided robust perioperative data do emerge, well-established continuous glucose monitoring technology could soon help transform the perioperative landscape.

The importance of intravenous glucose tolerance test glucose stimulus for the evaluation of insulin secretion.

For 100 years, the Intravenous glucose tolerance test (IVGTT) has been used extensively in researching the pathophysiology of diabetes mellitus and AIRg-the IVGTT-induced acute insulin response to the rapid rise in circulating glucose-is a key measure of insulin secretory capacity. For an effective evaluation of AIRg, IVGTT glucose loading should be adjusted for glucose distribution volume (gVOL) to provide an invariant, trend-free immediate rise in circulating glucose (ΔG0). Body weight-based glucose loads have been widely used but whether these achieve a trend-free ΔG0 does not appear to have been investigated. By analysing variation in AIRg, ΔG0 and gVOL with a range of IVGTT loads, both observed and simulated, we explored the hypothesis that there would be an optimum anthropometry-based IVGTT load calculation that, by achieving a trend-free ΔG0, would not compromise evaluation of AIRg as an index of beta cell function. Data derived from patient and research volunteer records for 3806 IVGTT glucose and insulin profiles. Among the non-obese, as gVOL rose, weight increased disproportionately rapidly. Consequently, the IVGTT glucose load needed for an invariant ΔG0 was progressively overestimated, accounting for 47% of variation in AIRg. Among the obese, ΔG0 was trend-free yet AIRg increased by 11.6% per unit body mass index, consistent with a more proportionate increase in weight with gVOL and a hyperinsulinaemic adaptation to adiposity-associated insulin resistance. Simulations further confirmed our hypothesis by demonstrating that a body surface area-based IVGTT load calculation could provide for a more generally invariant IVGTT ΔG0.

Developing an intervention to support dietary change for shift workers living with type 2 diabetes: A stakeholder consultation.

BACKGROUND: Shift workers, compared to day workers, are more likely to be diagnosed with type 2 diabetes (T2D). Currently, there is no tailored programme of dietary support available to either shift workers living with T2D or employers. METHODS: An intervention development consultation workshop was convened in June 2023 with the aim of evaluating potential interventions to identify those with a potential to take forward for further development. Findings from prior formative research into factors influencing dietary behaviour in shift workers with T2D were mapped to potential interventions addressing the barriers and enablers to healthy eating reported by shift workers with T2D. The findings of the Shift-Diabetes Study were presented in the context of the COM-B (Capability, Opportunity, Motivation, Behaviour) theoretical framework of behaviour change. Three interventions in turn were presented to attendees: (1) Educational resources and structured education, (2) Increasing availability and accessibility of food on a night shift and (3) Biofeedback and tailored advice. Seven workshop attendees were invited to express their thoughts, using the APEASE criteria (Affordability, Practicability, Effectiveness, Acceptability, Side-effects/Safety, Equity) to guide the discussion. The workshop was conducted online and recorded, and transcripts were thematically coded to the APEASE framework. RESULTS/CONCLUSIONS: The workshop highlighted the importance of multilevel interventions to support dietary behaviour change in this occupational group. Priority actions identified include (i) understanding barriers to 24/7 food availability, (ii) including shift workers in clinical diabetes studies and (iii) research to understand the effectiveness of continuous glucose monitoring in shift workers with T2D.

Switching from Intermittently Scanned Continuous Glucose Monitoring to Real-Time Continuous Glucose Monitoring with a Predictive Urgent Low Soon Alert Reduces Exposure to Hypoglycemia.

Differences in the effectiveness of real-time continuous glucose monitoring (rtCGM) and intermittently scanned continuous glucose monitoring (isCGM) in type 1 diabetes (T1D) are reported. The impact on percent time in range of switching from an isCGM with glucose threshold-based optional alerts only (FreeStyle Libre 2 [FSL2]) to an rtCGM (Dexcom G7) with an urgent low soon predictive alert was assessed, alongside other secondary outcomes including hemoglobin A1c (HbA1c) and other continuous glucose monitoring metrics. Adults with T1D using FSL2 were switched to Dexcom G7 for 12 weeks. HbA1c and continuous glucose data during FSL2 and Dexcom G7 use were compared. Data from 29 participants (aged 44.8 ± 16.5 years, 12 male and 17 female) were analyzed. After switching to rtCGM, participants spent less time in hypoglycemia below 3.9 mmol/L (70 mg/dL) (3.0% [1.0%, 5.0%] vs. 2.0% [1.0%, 3.0%], P = 0.006) and had higher percentage achievement of time below 3.9 mmol/L (70 mg/dL) of <4% (55.2% vs. 82.8%, P = 0.005). Coefficient of variation was lower (39.3 ± 6.6% vs. 37.2 ± 5.6%, P = 0.008). In conclusion, adults with T1D who switched from isCGM to rtCGM may benefit from reduced exposure to hypoglycemia and glycemic variability.

Increased glycemic variability in pregnant women with Roux-en-Y gastric bypass compared with sleeve gastrectomy.

INTRODUCTION: Bariatric surgery is associated with adverse pregnancy outcomes such as reduced birth weight and premature birth. One possible mechanism for this is increased glycemic variability (GV) which occurs after bariatric surgery. The objective of this study was to compare the effect of Roux-en-Y gastric bypass (RYGB) versus vertical sleeve gastrectomy (SG) on GV during pregnancy and to investigate the relationships of GV, type of bariatric surgery and maternal and neonatal outcomes. RESEARCH DESIGN AND METHODS: Fourteen pregnant women after RYGB and 14 after SG were investigated with continuous glucose monitoring in their second or third trimester in this observational study carried out as part of routine clinical care. RESULTS: Pregnant women with RYGB had similar mean interstitial glucose values but significantly increased indices of GV and a lower %time in range 3.9-7.8 mmol/L (70-140 mg/dL), compared with SG. CONCLUSIONS: Pregnant women who have undergone RYGB have greater GV during pregnancy compared with those who have undergone SG. Further research is needed to establish the relationship between GV and pregnancy outcomes to determine the preferred bariatric operation in women of reproductive age, and whether interventions to reduce GV might improve outcomes.

Exploration of the individual, social and environmental factors influencing dietary behaviour in shift workers with type 2 diabetes working in UK healthcare-The Shift-Diabetes Study: A qualitative study using the theoretical domains framework.

AIM: To identify factors influencing dietary behaviour in shift workers with type 2 diabetes (T2D) working in UK healthcare settings. METHODS: Semi-structured qualitative interviews based on the theoretical domains framework (TDF) were conducted with a convenience sample (n = 15) of shift workers (32-59 years) diagnosed with T2D who worked night shifts as part of a mixed shift schedule. The TDF was applied to analyse transcripts using a combined deductive framework and inductive thematic analysis approach. Identified influences were mapped to the behaviour change technique taxonomy to identify potential strategies to change dietary behaviour in this context. RESULTS: Key barriers to healthy dietary behaviours were access and cost of food available during night work (TDF domain: Environment Context and Resources). Factors identified as both enablers and barriers included: availability of staff facilities and time to take a break, (Environment Context and Resources), the physical impact of night work (Beliefs About Consequences), eating in response to stress or tiredness (Emotion), advance planning of meals/food and taking own food to work (Behavioural Regulation). Potential techniques to address these influences and improve dietary behaviour in this context include: meal planning templates, self-monitoring and biofeedback, and increasing accessibility and availability of healthier food choices during night shifts. CONCLUSIONS: The dietary behaviour of shift workers with T2D is influenced by interacting individual, socio-cultural and environmental factors. Intervention should focus on environmental restructuring and strategies that enable monitoring and meal planning.

A Randomized Controlled Trial Assessing the Impact of Continuous Glucose Monitoring with a Predictive Hypoglycemia Alert Function on Hypoglycemia in Physical Activity for People with Type 1 Diabetes (PACE).

Background: Uptake of exercise in people with type 1 diabetes (T1D) is low despite significant health benefits. Fear of hypoglycemia is the main barrier to exercise. Continuous glucose monitoring (CGM) with predictive alarms warning of impending hypoglycemia may improve self-management of diabetes around exercise. Aim: To assess the impact of Dexcom G6 real-time CGM system with a predictive hypoglycemia alert function on the frequency, duration, and severity of hypoglycemia occurring during and after regular (≥150 min/week) physical activity in people with T1D. Methods: After 10 days of blinded run-in (Baseline), CGM was unblinded and participants randomized 1:1 to have the "urgent low soon" (ULS) alert switched "on" or "off" for 40 days. Participants then switched alerts "off" or "on," respectively, for a further 40 days. Physical activity, and carbohydrate and insulin doses were recorded. Results: Twenty-four participants (8 men, 16 women) were randomized. There was no difference in change from baseline of hypoglycemia <3.0 and <3.9 mmol/L with the ULS on or off during the 24 h after exercise. With ULS alert "on" time spent below 2.8 mmol/L compared with baseline was significantly (P = 0.04) lower than with ULS "off" in the 24 h after exercise. In mixed effects regression, timing of the exercise and baseline HbA1c independently affected risk of hypoglycemia during exercise; exercise timing also affected hypoglycemia risk after exercise. Conclusion: A CGM device with an ULS alert reduces exposure to hypoglycemia below 2.8 mmol/L overall and in the 24 h after exercise compared with a threshold alert.

Hyperglycaemia aversion in type 1 diabetes: A grounded theory study.

OBJECTIVE: Very little is known about the circumstances under which hyperglycaemia aversion develops and is maintained. The present study aimed to identify psychological factors involved in the process of hyperglycaemia aversion and to understand how it affects people's self-management of type 1 diabetes. DESIGN: Qualitative, in-depth interviews were used. METHODS: A constructivist grounded theory study, using semi-structured participant interviews, was undertaken to build a theoretical model of the process of hyperglycaemia aversion. RESULTS: Eighteen participants were interviewed. Fifteen were considered hyperglycaemia averse and included in the analysis. A theoretical model was developed to describe and explain processes involved in hyperglycaemia aversion. Many participants held very high standards for themselves and often had a strong preference for control. While some participants described anxiety associated with higher blood glucose, the most proximal driver of their approach was self-criticism and frustration associated with not meeting their own high standards for blood glucose. A number of attentional processes and beliefs, mostly related to hypoglycaemia, maintained and reinforced their blood glucose preference. Diabetes technology served as an enabler, raiser of standards, and additional critical judge of participants' hyperglycaemia aversion. CONCLUSIONS: The trans-diagnostic concept of emotional over-control is used to understand the proposed model of processes of hyperglycaemia aversion. The present study offers new insight which will aid clinicians in identifying and supporting those who may be at risk of psychological distress and harm associated with a preference for avoidance of higher blood glucose levels.

A narrative commentary about interoperability in medical devices and data used in diabetes therapy from an academic EU/UK/US perspective.

People living with diabetes have many medical devices available to assist with disease management. A critical aspect that must be considered is how systems for continuous glucose monitoring and insulin pumps communicate with each other and how the data generated by these devices can be downloaded, integrated, presented and used. Not only is interoperability associated with practical challenges, but also devices must adhere to all aspects of regulatory and legal frameworks. Key issues around interoperability in terms of data ownership, privacy and the limitations of interoperability include where the responsibility/liability for device and data interoperability lies and the need for standard data-sharing protocols to allow the seamless integration of data from different sources. There is a need for standardised protocols for the open and transparent handling of data and secure integration of data into electronic health records. Here, we discuss the current status of interoperability in medical devices and data used in diabetes therapy, as well as regulatory and legal issues surrounding both device and data interoperability, focusing on Europe (including the UK) and the USA. We also discuss a potential future landscape in which a clear and transparent framework for interoperability and data handling also fulfils the needs of people living with diabetes and healthcare professionals.