Predicting glucose level with an adapted branch predictor

Blood glucose control is an essential goal for the patients who have Type‐1 diabetes (T1D). The prediction of the blood glucose levels for the next 30‐minute is crucial. If the predicted blood glucose level is in the critical ranges, and these predictions can be known in advance, then the patients can take the necessary cautions to prevent from it. In this article, we propose a modified fuzzy particle swarm optimization algorithm for the prediction of blood glucose levels of 30‐minute after the last measurement. We form the average and patient‐specific models to predict the blood glucose level of the patients. Both models are tested on two different datasets which contain patients with T1D. The experimental results are evaluated in terms of root mean squared error and Clarke error grid analysis metrics. The results indicate that our proposed modified algorithm is feasible to be applied to the prediction of blood glucose levels. In addition, this approach can assist patients with T1D for their blood glucose control.

Advances in exercise, physical activity, and diabetes mellitus.

Machine learning (ML) models provide more choices to patients with diabetes mellitus (DM) to more properly manage blood glucose (BG) levels. However, because of numerous types of ML algorithms, choosing an appropriate model is vitally important. In a systematic review and network meta-analysis, this study aimed to comprehensively assess the performance of ML models in predicting BG levels. In addition, we assessed ML models used to detect and predict adverse BG (hypoglycemia) events by calculating pooled estimates of sensitivity and specificity. PubMed, Embase, Web of Science, and Institute of Electrical and Electronics Engineers Explore databases were systematically searched for studies on predicting BG levels and predicting or detecting adverse BG events using ML models, from inception to November 2022. Studies that assessed the performance of different ML models in predicting or detecting BG levels or adverse BG events of patients with DM were included. Studies with no derivation or performance metrics of ML models were excluded. The Quality Assessment of Diagnostic Accuracy Studies tool was applied to assess the quality of included studies. Primary outcomes were the relative ranking of ML models for predicting BG levels in different prediction horizons (PHs) and pooled estimates of the sensitivity and specificity of ML models in detecting or predicting adverse BG events. In total, 46 eligible studies were included for meta-analysis. Regarding ML models for predicting BG levels, the means of the absolute root mean square error (RMSE) in a PH of 15, 30, 45, and 60 minutes were 18.88 (SD 19.71), 21.40 (SD 12.56), 21.27 (SD 5.17), and 30.01 (SD 7.23) mg/dL, respectively. The neural network model (NNM) showed the highest relative performance in different PHs. Furthermore, the pooled estimates of the positive likelihood ratio and the negative likelihood ratio of ML models were 8.3 (95% CI 5.7-12.0) and 0.31 (95% CI 0.22-0.44), respectively, for predicting hypoglycemia and 2.4 (95% CI 1.6-3.7) and 0.37 (95% CI 0.29-0.46), respectively, for detecting hypoglycemia. Statistically significant high heterogeneity was detected in all subgroups, with different sources of heterogeneity. For predicting precise BG levels, the RMSE increases with a rise in the PH, and the NNM shows the highest relative performance among all the ML models. Meanwhile, current ML models have sufficient ability to predict adverse BG events, while their ability to detect adverse BG events needs to be enhanced. PROSPERO CRD42022375250; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=375250.

Many factors affect blood glucose levels in type 1 diabetics, several of which vary largely both in magnitude and delay of the effect. Modern rapid-acting insulins generally have a peak time after 60–90 min, while carbohydrate intake can affect blood glucose levels more rapidly for high glycemic index foods, or slower for other carbohydrate sources. It is important to have good estimates of the development of glucose levels in the near future both for diabetic patients managing their insulin distribution manually, as well as for closed-loop systems making decisions about the distribution. Modern continuous glucose monitoring systems provide excellent sources of data to train machine learning models to predict future glucose levels. In this paper, we present an approach for predicting blood glucose levels for diabetics up to 1 h into the future. The approach is based on recurrent neural networks trained in an end-to-end fashion, requiring nothing but the glucose level history for the patient. Our approach obtains results that are comparable to the state of the art on the Ohio T1DM dataset for blood glucose level prediction. In addition to predicting the future glucose value, our model provides an estimate of its certainty, helping users to interpret the predicted levels. This is realized by training the recurrent neural network to parameterize a univariate Gaussian distribution over the output. The approach needs no feature engineering or data preprocessing and is computationally inexpensive. We evaluate our method using the standard root-mean-squared error (RMSE) metric, along with a blood glucose-specific metric called the surveillance error grid (SEG). We further study the properties of the distribution that is learned by the model, using experiments that determine the nature of the certainty estimate that the model is able to capture.

Diabetes is a chronic disease and one of the major public health problems worldwide. It is a multifactorial disease, caused by genetic factors and lifestyle habits. Brazil had ∼ 16.8 million individuals living with diabetes in 2019 and is expected to reach 26 million people by 2045. There are global increasing needs for the development of noninvasive diagnostic methods and use of mobile health, mainly in face of the pandemic caused by the coronavirus disease 2019 (COVID-19). For daily glycemic control, diabetic patients use a portable glucometer for glycemic self-monitoring and need to prick their fingertips three or more times a day, generating a huge discomfort throughout their lives. Our goal here is to present a review with very recent emerging studies in the field of noninvasive diagnosis and to emphasize that smartphone-based photoplethysmography (spPPG), powered by artificial intelligence, might be a trend to self-monitor blood glucose levels. In photoplethysmography, a light source travels through the tissue, interacts with the interstitium and with cells and molecules present in the blood. Reflection of light occurs as it passes through the biological tissues and a photodetector can capture these interactions. When using a smartphone, the built-in flashlight is a white light-emitting LED and the camera works as a photodetector. The higher the concentration of circulating glucose, the greater the absorbance and, consequently, the lesser the reflected light intensity will be. Due to these optical phenomena, the signal intensity captured will be inversely proportional to the blood glucose level. Furthermore, we highlight the microvascular changes in the progression of diabetes that can interfere in the signals captured by the photodetector using spPPG, due to the decrease of peripheral blood perfusion, which can be confused with high blood glucose levels. It is necessary to create strategies to filter or reduce the impact of these vascular changes in the blood glucose level analysis. Deep learning strategies can help the machine to solve these challenges, allowing an accurate blood glucose level and interstitial glucose prediction.

This work investigates the impact of patients' heterogeneity, including factors such as gender, age, and clinical conditions, on the performance of machine learning models in predicting blood glucose levels in individuals with Type 1 Diabetes. Using the T1DiabetesGranada dataset, various datasets were generated based on these heterogeneity factors. Three popular prediction models -a Linear model, an LSTM neural network, and a CNN- were applied to both the generated datasets and the complete dataset to measure prediction performance at a 30-minute prediction horizon. The preliminary results suggest that incorporating patient-specific heterogeneity factors generally improves prediction performance, highlighting the existence of bias in standard blood glucose level prediction models. Future research should explore whether these findings hold in other related datasets.

Diseases involving inflammation and oxidative stress can be exacerbated by high blood glucose levels. Due to tight metabolic regulation, safely reducing blood glucose can prove difficult. The ketogenic diet (KD) reduces absolute glucose and insulin, while increasing fatty acid oxidation, ketogenesis, and circulating levels of β-hydroxybutyrate (βHB), acetoacetate (AcAc), and acetone. Compliance to KD can be difficult, so alternative therapies that help reduce glucose levels are needed. Exogenous ketones provide an alternative method to elevate blood ketone levels without strict dietary requirements. In this study, we tested the changes in blood glucose and ketone (βHB) levels in response to acute, sub-chronic, and chronic administration of various ketogenic compounds in either a post-exercise or rested state. WAG/Rij (WR) rats, a rodent model of human absence epilepsy, GLUT1 deficiency syndrome mice (GLUT1D), and wild type Sprague Dawley rats (SPD) were assessed. Non-pathological animals were also assessed across different age ranges. Experimental groups included KD, standard diet (SD) supplemented with water (Control, C) or with exogenous ketones: 1, 3-butanediol (BD), βHB mineral salt (KS), KS with medium chain triglyceride/MCT (KSMCT), BD acetoacetate diester (KE), KE with MCT (KEMCT), and KE with KS (KEKS). In rested WR rats, the KE, KS, KSMCT groups had lower blood glucose level after 1 h of treatment, and in KE and KSMCT groups after 24 h. After exercise, the KE, KSMCT, KEKS, and KEMCT groups had lowered glucose levels after 1 h, and in the KEKS and KEMCT groups after 7 days, compared to control. In GLUT1D mice without exercise, only KE resulted in significantly lower glucose levels at week 2 and week 6 during a 10 weeks long chronic feeding study. In 4-month and 1-year-old SPD rats in the post-exercise trials, blood glucose was significantly lower in KD and KE, and in KEMCT groups, respectively. After seven days, the KSMCT group had the most significantly reduced blood glucose levels, compared to control. These results indicate that exogenous ketones were efficacious in reducing blood glucose levels within and outside the context of exercise in various rodent models of different ages, with and without pathology.

Type 1 Diabetes Mellitus

Closed-loop insulin delivery systems, which integrate continuous glucose monitoring (CGM) and algorithms that continuously guide insulin dosing, have been shown to improve glycaemic control. The ability to predict future glucose values can further optimize such devices. In this study, we used machine learning to train models in predicting future glucose levels based on prior CGM and accelerometry data. We used data from The Maastricht Study, an observational population-based cohort that comprises individuals with normal glucose metabolism, prediabetes, or type 2 diabetes. We included individuals who underwent >48h of CGM (n = 851), most of whom (n = 540) simultaneously wore an accelerometer to assess physical activity. A random subset of individuals was used to train models in predicting glucose levels at 15- and 60-minute intervals based on either CGM data or both CGM and accelerometer data. In the remaining individuals, model performance was evaluated with root-mean-square error (RMSE), Spearman's correlation coefficient (rho) and surveillance error grid. For a proof-of-concept translation, CGM-based prediction models were optimized and validated with the use of data from individuals with type 1 diabetes (OhioT1DM Dataset, n = 6). Models trained with CGM data were able to accurately predict glucose values at 15 (RMSE: 0.19mmol/L; rho: 0.96) and 60 minutes (RMSE: 0.59mmol/L, rho: 0.72). Model performance was comparable in individuals with type 2 diabetes. Incorporation of accelerometer data only slightly improved prediction. The error grid results indicated that model predictions were clinically safe (15 min: >99%, 60 min >98%). Our prediction models translated well to individuals with type 1 diabetes, which is reflected by high accuracy (RMSEs for 15 and 60 minutes of 0.43 and 1.73 mmol/L, respectively) and clinical safety (15 min: >99%, 60 min: >91%). Machine learning-based models are able to accurately and safely predict glucose values at 15- and 60-minute intervals based on CGM data only. Future research should further optimize the models for implementation in closed-loop insulin delivery systems.

61-Year-Old Man With Nausea and Vomiting

Both diabetic patients and health care practitioners often assume that the subjective symptoms of extreme low and high blood glucose (BG) levels are easily recognized. This study tested the accuracy of BG symptom beliefs in a group of 26 insulin-dependent diabetic patients. A within-subject, repeated measures design was used to identify symptoms related to low and high glucose levels for individual subjects. On three occasions over a 1-year period, subjects completed symptom checklists just prior to 40 consecutive self-measurements of BG (an average of four times per day for 10 days). At the end of the year, subjects reported which symptoms they believed were related to their own low and high glucose levels. After determining whether each checklist item was empirically related to glucose levels and believed to relate to glucose levels, symptom beliefs were categorized as hits, false alarms, misses, or correct rejections. Across subjects, the frequency of accurate beliefs (hits and correct rejections) was higher than the frequency of inaccurate beliefs (false alarms and misses). Symptom belief accuracy differed greatly for individual subjects, however, and every subject had at least one inaccurate belief. False alarm beliefs were the more common type of error. Female subjects' symptom beliefs yielded more hits, as well as more false alarms. Males missed more symptoms, especially low BG symptoms, than females. Symptom belief accuracy was greater if symptom-glucose relationships remained stable across time.

Evaluation of insulin sensitivity temporal prediction by using quantile regression combined with neural network model.

ALA16VAL-MnSOD gene polymorphism and stroke: Association with dyslipidemia and glucose levels

Autonomic nervous system (ANS) can maintain homeostasis through the coordination of different organs including heart. The change of blood glucose (BG) level can stimulate the ANS, which will lead to the variation of Electrocardiogram (ECG). Considering that the monitoring of different BG ranges is significant for diabetes care, in this paper, an ECG-based technique was proposed to achieve non-invasive monitoring with three BG ranges: low glucose level, moderate glucose level, and high glucose level. For this purpose, multiple experiments that included fasting tests and oral glucose tolerance tests were conducted, and the ECG signals from 21 adults were recorded continuously. Furthermore, an approach of fusing density-based spatial clustering of applications with noise and convolution neural networks (DBSCAN-CNN) was presented for ECG preprocessing of outliers and classification of BG ranges based ECG. Also, ECG's important information, which was related to different BG ranges, was graphically visualized. The result showed that the percentages of accurate classification were 87.94% in low glucose level, 69.36% in moderate glucose level, and 86.39% in high glucose level. Moreover, the visualization results revealed that the highlights of ECG for the different BG ranges were different. In addition, the sensitivity of prediabetes/diabetes screening based on ECG was up to 98.48%, and the specificity was 76.75%. Therefore, we conclude that the proposed approach for BG range monitoring and prediabetes/diabetes screening has potentials in practical applications.

We aimed to classify Japanese adults without diabetes into different categories based on the oral glucose tolerance test (OGTT) and characterize their insulin sensitivity and insulin secretion. The OGTT was performed on 1,085 Japanese individuals without diabetes (aged 20-64 years); blood glucose and insulin levels were measured at 0, 30-, 60-, 90-, and 120-min. Fasting blood chemistry, hematology, and urine were analyzed. The participants were classified into four categories based on the following: (A) 30 min post-load plasma glucose levels < 157 mg/dL and/or (B) 120 min post-load plasma glucose levels < 126 mg/dL and Matsuda index > 4.97. Category 1 satisfied both conditions, category 2 satisfied condition A but not B, category 3 satisfied condition B but not A, and category 4 satisfied neither condition. Overall, 46%, 21%, 13%, and 20% of the participants were classified into categories 1, 2, 3, and 4, respectively. Compared with category 1, the characteristics of the other categories were: 2, low insulin sensitivity and high blood glucose levels during the later period; 3, low insulin secretion and a rapid increase in blood glucose levels; and 4, combined characteristics of categories 2 and 3. Most blood test values besides glucose metabolism in category 4 were also worse than those in category 1. Categories 1 and 2 had a high proportion of females, whereas categories 3 and 4 had a low proportion. Japanese adults without diabetes are classified into four categories with different insulin sensitivities and insulin secretion using OGTT results. Each category has different characteristics of age and sex distribution and clinical values besides glucose metabolism.