Search

Aims/Introduction: We estimated the hazards of cardiovascular diseases (CVDs) and early all‐cause mortality in Korean adults according to the presence of recently diagnosed type 2 diabetes (type 2 diabetes for <5 years) and insulin use. Materials and Methods: We used the Korean National Health Insurance Service–National Sample Cohort database (2002–2015) for this longitudinal population‐based study. Among adults aged ≥40 years without baseline CVD, individuals without diabetes or with recently diagnosed type 2 diabetes were selected (N = 363,919). The hazard ratios (HRs) for myocardial infarction (MI), stroke, and all‐cause mortality during follow‐up were analyzed according to three groups categorized by the presence of type 2 diabetes and insulin use. Results: Within a mean 7.8 years, there were 5,275 MIs, 7,220 strokes, and 15,834 deaths. The hazards for outcomes were higher in the insulin‐treated type 2 diabetes group than in the non‐diabetes group [HR (95% CI): 2.344 (1.870–2.938) for MI, 2.420 (1.993–2.937) for stroke, and 3.037 (2.706–3.407) for death], higher in the non‐insulin‐treated type 2 diabetes group than in the non‐diabetes group [HR (95% CI): 1.284 (1.159–1.423) for MI, 1.435 (1.320–1.561) for stroke, and 1.135 (1.067–1.206) for death], and higher in the insulin‐treated type 2 diabetes group than in the non‐insulin‐treated type 2 diabetes group [HR (95% CI): 1.914 (1.502–2.441) for MI, 1.676 (1.363–2.060) for stroke, and 2.535 (2.232–2.880) for death]. Conclusions: Recently diagnosed type 2 diabetes patients showed increased risks of incident CVDs and premature mortality, and insulin‐treated group demonstrated an additional increase in the risks of these outcomes in adults with recently diagnosed type 2 diabetes, suggesting the need for intensified cardio‐protective interventions for adults with insulin‐treated type 2 diabetes. [ABSTRACT FROM AUTHOR]

Aims/Introduction: We aimed to determine whether mean and visit‐to‐visit glycated hemoglobin (HbA1c) variability independently increase the incidence of non‐alcoholic fatty liver disease (NAFLD) across the diabetic continuum from normal glucose tolerance (NGT) to established diabetes. Materials and Methods: In a longitudinal cohort study, 21,123 participants underwent five or more annual health screening checkups. Participants were categorized into diabetes (n = 1,635), prediabetes (n = 6,650) and NGT (n = 12,838) groups. Mean, standard deviation (SD) and coefficient of variation data on HbA1c were obtained from three consecutive measurements. The associations between those data and incident NAFLD were analyzed using Cox regressions. Results: Over a median follow‐up period of 57 months, 3,860 (18.3%) participants developed NAFLD. The risk of NAFLD increased continuously, with the mean HbA1c beginning at 4.9%, even in the NGT group. We found a significant association between increasing HbA1c variability and incident NAFLD (coefficient of variation, adjusted hazard ratio 1.14, 95% confidence interval 1.01–1.29; standard deviation, adjusted hazard ratio 1.19, 95% confidence interval 1.05–1.36) in the diabetes group, but not in the NGT or prediabetes group. Consistent findings were observed when NAFLD patients with a low possibility of fibrosis were excluded. The association between the coefficient of variation of HbA1c and incident NAFLD in the diabetes group was significant only in those with an increasing trend of post‐baseline HbA1c (adjusted hazard ratio 1.24, 95% confidence interval 1.01–1.52). Conclusions: Increased mean HbA1c levels elevated the risk of incident NAFLD, even with NGT. Increases in visit‐to‐visit variability of HbA1c independently elevated the risk of incident NAFLD, but only in the diabetes group. [ABSTRACT FROM AUTHOR]

• We propose a GAN-based data augmentation to solve data imbalance on hypoglycemia prediction problem. • With a large-scale CGM dataset, we develop conditional GAN and conditional Wasserstein GAN. • We compare the performance of various predictive models with several data augmentation methods. • In most cases, the proposed data augmentation method improves the performance of hypoglycemia prediction. Hypoglycemia is one of the major barriers for intensive insulin treatment to achieve optimal glycemic control for people with diabetes. Accurate prediction of hypoglycemia became an important factor for advancing insulin therapy, and thus numerous studies have proposed data-driven models. However, the data-driven models still suffer from performance degradation due to severe data imbalance between hypoglycemia and non-hypoglycemia. To overcome this problem, we propose a generative adversarial network (GAN) based data augmentation method, generating realistic continuous glucose monitoring (CGM) time series labeled hypoglycemia. Having acquired a large-scale CGM time series dataset, we compared the performance of various models before and after five data augmentation methods. The GAN-based data augmentation method improved the hypoglycemia prediction performance when combined with ML models and we found that the data augmentation method was comparable to conventional data augmentation method. Through visualization, it was found that successfully generated CGM time series satisfied a given condition, and the generated CGM time series were visually separated according to the given condition in an embedding space. These results suggest that GAN-based data augmentation is a promising approach for solving the severe data imbalance of hypoglycemia prediction. We believe that the combination of more accurate hypoglycemia prediction models and intensive insulin therapy will result in better glycemic control for people with diabetes. [ABSTRACT FROM AUTHOR]

Objective: The automation of insulin treatment is the most challenge aspect of glucose management for type 1 diabetes owing to unexpected exogenous events (e.g., meal intake). In this article, we propose a novel reinforcement learning (RL) based artificial intelligence (AI) algorithm for a fully automated artificial pancreas (AP) system. Methods: A bioinspired RL designing method was developed for automated insulin infusion. This method has reward functions that imply the temporal homeostatic objective and discount factors that reflect an individual specific pharmacological characteristic. The proposed method was applied to a training method using an RL algorithm and was evaluated in virtual patients from the FDA approved UVA/Padova simulator with unannounced meal intakes. Results: For a single-meal experiment with preprandial fasting, the trained policy demonstrated fully automated regulation in both the basal and postprandial phases. In the in silico trial with a variation of insulin sensitivity and dawn phenomenon, the policy achieved a mean glucose of 124.72 mg/dL and percentage time in the normal range of 89.56%. The layer-wise relevance propagation provides interpretable information on AI-driven decision for robustness to sensor noise, automated postprandial regulation, and insulin stacking avoidance. Conclusion: The AP algorithm based on the bioinspired RL approach enables fully automated blood glucose control with unannounced meal intake. Significance: The proposed framework can be extended to other drug-based treatments for systems with significant uncertainties. [ABSTRACT FROM AUTHOR]

Background: A Personal Health Record (PHR) is an online application that allows patients to access, manage, and share their health data. PHRs not only enhance shared decision making with healthcare providers, but also enable remote monitoring and at-home-collection of detailed data. The benefits of PHRs can be maximized in insulin dose adjustment for patients starting or intensifying insulin regimens, as frequent self-monitoring of glucose, self-adjustment of insulin dose, and precise at-home data collection during the visit-to-visit period are important for glycemic control. The aim of this study is to examine the efficacy and safety of insulin dose adjustment based on a smartphone PHR application in patients with diabetes mellitus (DM) and to confirm the validity and stability of an information and communication technology (ICT)-based centralized clinical trial monitoring system.Methods: This is a 24-week, open-label, randomized, multi-center trial. There are three follow-up measures: baseline, post-intervention at week 12, and at week 24. Subjects diagnosed with type 1 DM, type 2 DM, and/or post-transplant DM who initiate basal insulin or intensify their insulin regimen to a basal-bolus regimen are included. After education on insulin dose titration and prevention for hypoglycemia and a 1-week acclimation period, subjects are randomized in a 1:1 ratio to either an ICT-based intervention group or a conventional intervention group. Subjects in the conventional intervention group will save and send their health information to the server via a PHR application, whereas those in ICT-based intervention group will receive additional algorithm-based feedback messages. The health information includes level of blood glucose, insulin dose, details on hypoglycemia, food diary, and step count. The primary outcome will be the proportion of patients who reach an optimal insulin dose within 12 weeks of study enrollment, without severe hypoglycemia or unscheduled clinic visits.Discussion: This clinical trial will reveal whether insulin dose adjustment based on a smartphone PHR application can facilitate the optimization of insulin doses in patients with DM. In addition, the process evaluation will provide information about the validity and stability of the ICT-based centralized clinical trial monitoring system in this research field.Trial Registration: Clinicaltrials.gov NCT 03112343 . Registered on 12 April 2017. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Diabetes is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Aims/Introduction Small dense low-density lipoprotein (sd LDL) has been suggested to be a potential risk factor for cardiovascular diseases ( CVD). Materials and Methods We carried out a prospective nested case-control study in the Korean Health and Genome Study. Participants were men and women aged 40-69 years who developed CVD ( n = 313), and were matched by age and sex to controls who remained free of CVD ( n = 313) during the 8-years follow-up period (from 2001 to 2009). LDL subfractions were analyzed in frozen samples collected from the 626 participants using polyacrylamide tube gel electrophoresis. Results Patients with CVD had a significantly higher glycated hemoglobin level compared with the controls (5.72 vs 5.56). The proportion of patients with diabetes mellitus ( DM) was higher in those who developed CVD during follow up (8.0% vs 1.9%). The frequency of CVD according to each tertile of LDL particle size and the number of metabolic syndrome components did not differ significantly. In the multivariate analysis, DM (odds ratio 4.244, 95% confidence interval 1.693-10.640, P = 0.002) was the only independent predictive factor of CVD. LDL particle size was not associated with the risk for future CVD. Conclusions Small dense LDL might not be a significant predictor of CVD in this Korean community-based prospective cohort study. [ABSTRACT FROM AUTHOR]

Simple Summary: The present nationwide population-based longitudinal cohort study showed that diabetes was associated with an increased risk of hematologic malignancies independent of obesity. The risk of NHL increased according to the progression of dysglycemia towards a longer diabetes duration, while HL did not. There have been conflicting results regarding the association between diabetes and the risk of hematologic malignancies, and its interaction with obesity is unknown. This study determined the risk of hematologic malignancies according to the glycemic status in a population-based study involving health screening 9,774,625 participants. The baseline glycemic status of the participants was categorized into no diabetes, impaired fasting glucose (IFG), newly detected diabetes, diabetes duration <5 years, and diabetes duration ≥5 year groups. The risks of overall and specific hematologic malignancies were estimated using a Cox regression analysis. During a median follow up of 7.3 years, 14,733 hematologic malignancies developed. The adjusted hazard ratio (aHR) for the risk of all the hematologic malignancies was 0.99 (95% confidence interval (CI) 0.95–1.02) for IFG, 0.99 (95% CI 0.91–1.08) for newly detected diabetes, 1.03 (95% CI 0.96–1.11) for diabetes duration <5 years, and 1.11 (95% CI 1.03, 1.20) for diabetes duration ≥5 year groups. The association was independent from obesity. The risk of non-Hodgkin's lymphoma (NHL) increased according to the progression of dysglycemia towards a longer diabetes duration, while Hodgkin's lymphoma did not. This study in Korea demonstrated diabetes to be associated with an increased risk of hematologic malignancies independent of obesity. The NHL risk increased with the diabetes duration. [ABSTRACT FROM AUTHOR]

Aims: We examined the association between changes in serum calcium levels with the incidence of type 2 diabetes mellitus (T2DM) in apparently healthy South Korean subjects.Methods: A retrospective longitudinal analysis was conducted with subjects who had participated in comprehensive health check-ups at least four times over a 7-year period (between 2006 and 2012). In total, 23,121 subjects were categorized into tertiles based on changes in their albumin-adjusted serum calcium levels. Multivariate Cox regression models were fitted to assess the association between changes in serum calcium levels during follow-up and the relative risk of diabetes incidence.Results: After a median follow-up of 57.4months, 1,929 (8.3%) new cases of T2DM occurred. Simple linear regression analysis showed serum calcium level changes correlated positively with changes in HbA1c and fasting plasma glucose (FPG) levels (B=5.72, p<0.001 for FPG; B=0.13, p<0.001 for HbA1c). An increase in albumin-adjusted serum calcium levels during follow-up was related to an increased risk of T2DM. After adjustment for potential confounders, the risk of T2DM was 1.6 times greater for subjects whose albumin-adjusted serum calcium levels were in the highest change tertile during follow-up than for subjects whose levels were in the lowest tertile (HR 1.65, 95% CI 1.44-1.88, P<0.001).Conclusions: The elevation of albumin-adjusted serum calcium levels was associated with an increased risk of T2DM, independent of baseline glycemic status. [ABSTRACT FROM AUTHOR]

Introduction Glycated albumin (GA) reflects short-term status of glycemic control. We suggest a GA cut-off value to diagnose pre-diabetes and diabetes in Korean adults. In addition, we compared the performance of GA for the diagnosis of diabetes with that of glycated hemoglobin (A1c). Materials and methods A total of 852 subjects (498 males, 354 females) aged 20 to 83 years (mean: 52.5 years) were enrolled. A 75-g oral glucose tolerance test (OGTT) was performed and A1c and GA were measured. Results In these enrolled subjects, 88% have glucose intolerance status (pre-diabetes or diabetes). The GA concentrations corresponding to fasting plasma glucose (FPG) of 7.0 mmol/l, 2-h plasma glucose during OGTT (PPG2) ≥ 11.1 mmol/l, and A1c ≥ 6.5% were 14.6%, 13.7%, and 14.7%, respectively. A meta-analysis of three GA cutoffs revealed a GA cutoff for diabetes of 14.3%. When A1c is used in combination with FPG, the sensitivity and specificity for the diagnosis of OGTT-based diabetes were 72.16% (95% CI: 66.6–72.2) and 96.4% (95% CI: 94.4–97.7), respectively. With the newly developed GA cutoff of 14.3%, GA combined with FPG resulted in a sensitivity and specificity of 77.5% (95% CI: 72.17–82.0) and 89.9% (95% CI: 87.1–92.2), respectively. Conclusions A GA cutoff of > 14.3% is optimal for the diagnosis of diabetes in Korean adults. The measurement of FPG and GA may detect diabetes earlier than the measurement of FPG and A1c. [ABSTRACT FROM AUTHOR]

• We proposed a personalized blood glucose (BG) level prediction model with a fine-tuning strategy and demonstrated its efficacy on large datasets including three types of diabetes (type 1 diabetes, type 2 diabetes, and gestational diabetes). • The fine-tuned convolutional neural network (CNN) showed the performance of the general CNN in most cases and outperformed the scratch CNN. • We analyzed all cases of four predictive patterns and found that the input BG level trend and the BG level at the time of prediction were important in determining the future BG level trend. • We believe that our method's method and results will be useful for building the personalized model and analyzing its predictions. The accurate prediction of blood glucose (BG) level is still a challenge for diabetes management. This is due to various factors such as diet, personal physiological characteristics, stress, and activities influence changes in BG level. To develop an accurate BG level predictive model, we propose a personalized model based on a convolutional neural network (CNN) with a fine-tuning strategy. We utilized continuous glucose monitoring (CGM) datasets from 1052 professional CGM sessions and split them into three groups according to type 1, type 2, and gestational diabetes mellitus (T1DM, T2DM, and GDM, respectively). During the preprocessing, only CGM data points were utilized, and future BG levels of four different prediction horizons (PHs, 15, 30, 45, and 60 min) were used as output. In training, we trained a general CNN and a multi-output random forest regressor using a hold-out method for each group. Next, we developed two personalized models: (1) by fine-tuning the general CNN on partial sample points of each CGM dataset, and (2) by learning a CNN from scratch on the points. For all groups, the fine-tuned CNN showed the lowest average root mean squared error, average mean absolute percentage error, highest average time gain (PH = 15 and 60 min in T1DM) and highest percentage in region A of Clarke error grid analysis at all PHs. In the performance comparison between the fine-tuned CNN and other models, we found that the fine-tuned CNN improved the performance of the general CNN in most cases and outperformed the scratch CNN at all PHs in all groups, making the fine-tuning strategy was useful for accurate BG level prediction. We analyzed all cases of four predictive patterns in each group, and found that the input BG level trend and the BG level at the time of prediction were related to the future BG level trend. We demonstrated the efficacy of the fine-tuning method in a large number of CGM datasets and analyzed the four predictive patterns. Therefore, we believe that the proposed method will significantly contribute to the development of an accurate personalized model and the analysis for its predictions. [ABSTRACT FROM AUTHOR]