An official website of the United States government
The management of blood glucose levels is critical in the care of Type 1 diabetes subjects. In extremes, high or low levels of blood glucose are fatal. To avoid such adverse events, there is the development and adoption of wearable technologies that continuously monitor blood glucose and administer insulin. This technology allows subjects to easily track their blood glucose levels with early intervention without the need for hospital visits. The data collected from these sensors is an excellent candidate for the application of machine learning algorithms to learn patterns and predict future values of blood glucose levels. In this study, we developed artificial neural network algorithms based on the OhioT1DM training dataset that contains data on 12 subjects. The dataset contains features such as subject identifiers, continuous glucose monitoring data obtained in 5 minutes intervals, insulin infusion rate, etc. We developed individual models, including LSTM, BiLSTM, Convolutional LSTMs, TCN, and sequence-to-sequence models. We also developed transfer learning models based on the most important features of the data, as identified by a gradient boosting algorithm. These models were evaluated on the OhioT1DM test dataset that contains 6 unique subject’s data. The model with the lowest RMSE values for the 30- and 60-minutes was selected as the best performing model. Our result shows that sequence-to-sequence BiLSTM performed better than the other models. This work demonstrates the potential of artificial neural networks algorithms in the management of Type 1 diabetes.
more »
« less
This content will become publicly available on March 17, 2027
A wireless sweat sensing with a pH-based correlation model for continuous glucose monitoring and diabetes management during exercise
In situ monitoring of sweat glucose during exercise can provide a real-time and continuous assessment of blood glucose dynamics. However, the relatively poor correlation between sweat and blood glucose concentrations during exercise makes it challenging for blood glucose management (BGM) during exercise therapy for diabetes, along with training for athletes and fitness enthusiasts. This work presents a flexible wireless sweat glucose and pH sensing platform integrated with a pH-based correlation model to accurately predict the continuous changes in blood glucose. The pH-based correlation model calibrates enzyme activity changes in glucose oxidase and accounts for the effects of sweat dilution and filtering during paracellular transport of glucose from interstitial fluid and plasma to sweat during exercise. The correlation model has been validated in both healthy individuals and diabetic patients, revealing distinct blood glucose dynamic patterns between the two cohorts. The observed different glucose fluctuations after the intake of various nutritive foods further facilitate the management of diabetes and allow for the identification of hypo-/hyperglycemic risks during training or fitness exercise. The exercise-based device platform combines continuous blood glucose monitoring with diabetes management through effective treatment evaluation and can also provide early prevention for the at-risk population and reduce or even reverse diabetes.
more »
« less
- PAR ID:
- 10694985
- Publisher / Repository:
- The National Academy of Sciences (NAS)
- Date Published:
- Journal Name:
- Proceedings of the National Academy of Sciences
- Volume:
- 123
- Issue:
- 11
- ISSN:
- 0027-8424
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Although increasing efforts have been devoted to the development of non‐invasive wearable electrochemical sweat sensors for monitoring physiological and metabolic information, most of them still suffer from poor stability and specificity over time and fluctuating temperatures. This study reports the design and fabrication of a long‐term stable and highly sensitive flexible electrochemical sensor based on nanocomposite‐modified porous graphene by facile laser treatment for detecting biomarkers such as glucose in sweat. The laser‐reduced and patterned stable conductive nanocomposite on the porous graphene electrode provides the resulting glucose sensor with an excellent sensitivity of 1317.69 µA mm−1cm−2and an ultra‐low limit of detection of 0.079 µm. The sensor can also detect pH and exhibit extraordinary stability to maintain more than 91% sensitivity over 21 days in ambient conditions. Taken together with a temperature sensor based on the same material system, the dual glucose and pH sensor integrated with a flexible microfluidic sweat sampling network further results in accurate continuous on‐body glucose detection calibrated by the simultaneously measured pH and temperature. The low‐cost, highly sensitive, and long‐term stable platform could facilitate the early identification and continuous monitoring of different biomarkers for non‐invasive disease diagnosis and treatment evaluation.more » « less
-
Abstract Background Blood glucose (BG) management is crucial for type-1 diabetes patients resulting in the necessity of reliable artificial pancreas or insulin infusion systems. In recent years, deep learning techniques have been utilized for a more accurate BG level prediction system. However, continuous glucose monitoring (CGM) readings are susceptible to sensor errors. As a result, inaccurate CGM readings would affect BG prediction and make it unreliable, even if the most optimal machine learning model is used. Methods In this work, we propose a novel approach to predicting blood glucose level with a stacked Long short-term memory (LSTM) based deep recurrent neural network (RNN) model considering sensor fault. We use the Kalman smoothing technique for the correction of the inaccurate CGM readings due to sensor error. Results For the OhioT1DM (2018) dataset, containing eight weeks’ data from six different patients, we achieve an average RMSE of 6.45 and 17.24 mg/dl for 30 min and 60 min of prediction horizon (PH), respectively. Conclusions To the best of our knowledge, this is the leading average prediction accuracy for the ohioT1DM dataset. Different physiological information, e.g., Kalman smoothed CGM data, carbohydrates from the meal, bolus insulin, and cumulative step counts in a fixed time interval, are crafted to represent meaningful features used as input to the model. The goal of our approach is to lower the difference between the predicted CGM values and the fingerstick blood glucose readings—the ground truth. Our results indicate that the proposed approach is feasible for more reliable BG forecasting that might improve the performance of the artificial pancreas and insulin infusion system for T1D diabetes management.more » « less
-
Blood glucose monitoring is an essential aspect of disease management for individuals with diabetes. Unfortunately, traditional methods require collecting a blood sample and thus are invasive and inconvenient. Recent developments in minimally invasive continuous glucose monitors have provided a more convenient alternative for people with diabetes to track their glucose levels 24/7. Despite this progress, many challenges remain to establish a noninvasive monitoring technique that works accurately and reliably in the wild. This review encompasses the current advancements in noninvasive glucose sensing technology in vivo, delves into the common challenges faced by these systems, and offers an insightful outlook on existing and future solutions.more » « less
-
Continuous lipid profiling could reveal dynamic lipid fluctuations underlying cardiometabolic health, yet existing enzymatic sensing strategies cannot sustain the multi-enzyme reactions required for continuous lipid detection, particularly when non-regenerable cofactors are consumed. Here we introduce a wearable epidermal lipid profiler for non-invasive, continuous monitoring of cholesterol and triglycerides using dual-step and cascading enzymatic reactions supported by the continuous refreshment of adenosine triphosphate (ATP). The system integrates a polymeric cofactor-releasing module composed of polypyrrole–ATP nanostructures that provide controlled, sweat-triggered ATP replenishment for long-term biosensing. We demonstrate real-time lipid tracking during dietary challenges and habitual activities, capturing distinct postprandial responses to different macronutrient compositions. In a preclinical study, simultaneous sweat and blood analyses coupled with causal machine learning revealed key physiological confounders and established individualized mappings between sweat and blood lipids. This cofactor-refreshing platform extends continuous monitoring to complex enzymatic systems, establishing a foundation for non-invasive, longitudinal cardiometabolic health assessment.more » « less
