Accurate prediction of blood glucose variations in type 2 diabetes (T2D) will facilitate better glycemic control and decrease the occurrence of hypoglycemic episodes as well as the morbidity and mortality associated with T2D, hence increasing the quality of life of patients. Owing to the complexity of the blood glucose dynamics, it is difficult to design accurate predictive models in every circumstance, i.e., hypo/normo/hyperglycemic events. We developed deep-learning methods to predict patient-specific blood glucose during various time horizons in the immediate future using patient-specific every 30-min long glucose measurements by the continuous glucose monitoring (CGM) to predict future glucose levels in 5 min to 1 h. In general, the major challenges to address are (1) the dataset of each patient is often too small to train a patient-specific deep-learning model, and (2) the dataset is usually highly imbalanced given that hypo- and hyperglycemic episodes are usually much less common than normoglycemia. We tackle these two challenges using transfer learning and data augmentation, respectively. We systematically examined three neural network architectures, different loss functions, four transfer-learning strategies, and four data augmentation techniques, including mixup and generative models. Taken together, utilizing these methodologies we achieved over 95% prediction accuracy and 90% sensitivity for a time period within the clinically useful 1 h prediction horizon that would allow a patient to react and correct either hypoglycemia and/or hyperglycemia. We have also demonstrated that the same network architecture and transfer-learning methods perform well for the type 1 diabetes OhioT1DM public dataset.
Background This study aimed at assessing the long-term effects of intranasal insulin (INI) on cognition and gait in older people with and without type 2 diabetes mellitus (T2DM). Methods Phase 2 randomized, double-blinded trial consisted of 24 week treatment with 40 IU of INI (Novolin ® R, off-label use) or placebo (sterile saline) once daily and 24 week follow-up. Primary outcomes were cognition, normal (NW), and dual-task (DTW) walking speeds. Of 244 randomized, 223 completed baseline (51 DM-INI, 55 DM-Placebo, 58 Control-INI, 59 Control-Placebo; 109 female, 65.8 ± 9.1; 50–85 years old); 174 completed treatment (84 DM, 90 Controls); 156 completed follow-up (69 DM). Results DM-INI had faster NW (~ 7 cm/s; p = 0.025) and DTW on-treatment ( p = 0.007; p = 0.812 adjusted for baseline difference) than DM-Placebo. Control-INI had better executive functioning on-treatment ( p = 0.008) and post-treatment ( p = 0.007) and verbal memory post-treatment ( p = 0.004) than Control-Placebo. DM-INI increased cerebral blood flow in medio-prefrontal cortex ( p < 0.001) on MRI. Better vasoreactivity was associated with faster DTW ( p < 0.008). In DM-INI, plasma insulin ( p = 0.006) and HOMA-IR ( p < 0.013) decreased post-treatment. Overall INI effect demonstrated faster walking ( p = 0.002) and better executive function ( p = 0.002) and verbal memory ( p = 0.02) (combined DM-INI and Control-INI cohort, hemoglobin A1c-adjusted). INI was not associated with serious adverse events, hypoglycemic episodes, or weight gain. Conclusion There is evidence for positive INI effects on cognition and gait. INI-treated T2DM participants walked faster, showed increased cerebral blood flow and decreased plasma insulin, while controls improved executive functioning and verbal memory. The MemAID trial provides proof-of-concept for preliminary safety and efficacy and supports future evaluation of INI role to treat T2DM and age-related functional decline. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1007/s00415-022-11119-6.
Introduction: Glycemic variability (GV) has been associated with worse prognosis in critically ill patients. We sought to evaluate the potential association between GV indices and clinical outcomes in acute stroke patients. Methods: Consecutive diabetic and nondiabetic, acute ischemic or hemorrhagic stroke patients underwent regular, standard-of-care finger-prick measurements and continuous glucose monitoring (CGM) for up to 96 h. Thirteen GV indices were obtained from CGM data. Clinical outcomes during hospitalization and follow-up period (90 days) were recorded. Hypoglycemic episodes disclosed by CGM but missed by finger-prick measurements were also documented. Results: A total of 62 acute stroke patients [48 ischemic and 14 hemorrhagic, median NIHSS score: 9 (IQR: 3–16) points, mean age: 65 ± 10 years, women: 47%, nondiabetic: 79%] were enrolled. GV expressed by higher mean absolute glucose (MAG) values was associated with a lower likelihood of neurological improvement during hospitalization before and after adjusting for potential confounders (OR: 0.135, 95% CI: 0.024–0.751, p = 0.022). There was no association of GV indices with 3-month clinical outcomes. During CGM recording, 32 hypoglycemic episodes were detected in 17 nondiabetic patients. None of these episodes were identified by the periodic blood glucose measurements and therefore they were not treated. Conclusions: Greater GV of acute stroke patients may be related to lower odds of neurological improvement during hospitalization. No association was disclosed between GV indices and 3-month clinical outcomes.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.