It is presently unclear whether glycemic variability is associated with diabetic cardiovascular autonomic neuropathy (CAN). The aim of this study was to examine whether short- and/or long-term glycemic variability (GV) contribute to CAN. A total of 110 patients with type 2 diabetes who underwent three-day continuous glucose monitoring (CGM) completed five standardized autonomic neuropathy tests. Short-term GV was measured by the standard deviation (SD), coefficient of variation (CV) of glucose, and the mean amplitude of glycemic excursions (MAGE) in CGM. HbA1c variability was calculated from the intrapersonal SD, adjusted SD, and CV of serial HbA1c over 2-year period. CAN was defined as the presence of at least two abnormal parasympathetic function tests. The severity of CAN was evaluated by total scores of five autonomic function tests. In univariate analysis, not only SD and CV in CGM but also all parameters of HbA1c variability were significantly higher in the patients with CAN (n = 47, 42.7 %) than in those without CAN. In multivariate analysis, CV (Odds ratio [OR] 1.07, 95 % confidence interval [CI] 1.01–1.13; p = 0.033), but neither SD nor MAGE in CGM, independently correlated with the presence of CAN. All parameters of HbA1c variability, such as SD of HbA1c (OR 12.10 [95 % CI 2.29–63.94], p = 0.003), adjusted SD of HbA1c (OR 17.02 [95 % CI 2.66–108.86], p = 0.003), and log CV of HbA1c (OR 24.00 [95 % CI 3.09–186.48], p = 0.002), were significantly associated with the presence of CAN. The patients with higher HbA1c variability had an increased risk of advanced CAN. CV in CGM and all parameters of HbA1c variability were independently associated with the presence of CAN in patients with inadequately controlled type 2 diabetes requiring CGM.

Skeletal muscle mass was negatively associated with metabolic syndrome prevalence in previous cross-sectional studies. The aim of this study was to investigate the impact of baseline skeletal muscle mass and changes in skeletal muscle mass over time on the development of metabolic syndrome in a large population-based 7-year cohort study. A total of 14,830 and 11,639 individuals who underwent health examinations at the Health Promotion Center at Samsung Medical Center, Seoul, Korea were included in the analyses of baseline skeletal muscle mass and those changes from baseline over 1 year, respectively. Skeletal muscle mass was estimated by bioelectrical impedance analysis and was presented as a skeletal muscle mass index (SMI), a body weight-adjusted appendicular skeletal muscle mass value. Using Cox regression models, hazard ratio for developing metabolic syndrome associated with SMI values at baseline or changes of SMI over a year was analyzed. During 7 years of follow-up, 20.1% of subjects developed metabolic syndrome. Compared to the lowest sex-specific SMI tertile at baseline, the highest sex-specific SMI tertile showed a significant inverse association with metabolic syndrome risk (adjusted hazard ratio [AHR] = 0.61, 95% confidence interval [CI] 0.54–0.68). Furthermore, compared with SMI changes < 0% over a year, multivariate-AHRs for metabolic syndrome development were 0.87 (95% CI 0.78–0.97) for 0–1% changes and 0.67 (0.56–0.79) for > 1% changes in SMI over 1 year after additionally adjusting for baseline SMI and glycometabolic parameters. An increase in relative skeletal muscle mass over time has a potential preventive effect on developing metabolic syndrome, independently of baseline skeletal muscle mass and glycometabolic parameters.

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.