An official website of the United States government
Neural networks present a useful framework for learning complex dynamics, and are increasingly being considered as components to closed loop predictive control algorithms. However, if they are to be utilized in such safety-critical advisory settings, they must be provably "conformant" to the governing scientific (biological, chemical, physical) laws which underlie the modeled process. Unfortunately, this is not easily guaranteed as neural network models are prone to learn patterns which are artifacts of the conditions under which the training data is collected, which may not necessarily conform to underlying physiological laws. In this work, we utilize a formal range-propagation based approach for checking whether neural network models for predicting future blood glucose levels of individuals with type-1 diabetes are monotonic in terms of their insulin inputs. These networks are increasingly part of closed loop predictive control algorithms for "artificial pancreas" devices which automate control of insulin delivery for individuals with type-1 diabetes. Our approach considers a key property that blood glucose levels must be monotonically decreasing with increasing insulin inputs to the model. Multiple representative neural network models for blood glucose prediction are trained and tested on real patient data, and conformance is tested through our verification approach. We observe that standard approaches to training networks result in models which violate the core relationship between insulin inputs and glucose levels, despite having high prediction accuracy. We propose an approach that can learn conformant models without much loss in accuracy.
more »
« less
Robust Data-Driven Control of Artificial Pancreas Systems Using Neural Networks
In this paper, we provide an approach to data-driven control for artificial pancreas systems by learning neural network models of human insulin-glucose physiology from available patient data and using a mixed integer optimization approach to control blood glucose levels in real-time using the inferred models. First, our approach learns neural networks to predict the future blood glucose values from given data on insulin infusion and their resulting effects on blood glucose levels. However, to provide guarantees on the resulting model, we use quantile regression to fit multiple neural networks that predict upper and lower quantiles of the future blood glucose levels, in addition to the mean. Using the inferred set of neural networks, we formulate a model-predictive control scheme that adjusts both basal and bolus insulin delivery to ensure that the risk of harmful hypoglycemia and hyperglycemia are bounded using the quantile models while the mean prediction stays as close as possible to the desired target. We discuss how this scheme can handle disturbances from large unannounced meals as well as infeasibilities that result from situations where the uncertainties in future glucose predictions are too high. We experimentally evaluate this approach on data obtained from a set of 17 patients over a course of 40 nights per patient. Furthermore, we also test our approach using neural networks obtained from virtual patient models available through the UVA-Padova simulator for type-1 diabetes.
more »
« less
- PAR ID:
- 10098747
- Date Published:
- Journal Name:
- Lecture notes in computer science
- Volume:
- 11095
- ISSN:
- 0302-9743
- Page Range / eLocation ID:
- 183-202
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
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 study investigates the development of a supramolecular peptide amphiphile (PA) material functionalized with phenylboronic acid (PBA) for glucose-responsive glucagon delivery. The PA-PBA system self-assembles into nanofibrillar hydrogels in the presence of physiological glucose levels, resulting in stable hydrogels capable of releasing glucagon under hypoglycemic conditions. Glucose responsiveness is driven by reversible binding between PBA and glucose, which modulates the electrostatic interactions necessary for hydrogel formation and dissolution. Through comprehensive in vitro characterization, including circular dichroism, zeta potential measurements, and rheological assessments, the PA-PBA system is found to exhibit glucose-dependent assembly, enabling controlled glucagon release that is inversely related to glucose concentration. Glucagon release is accelerated under low glucose conditions, simulating a hypoglycemic state, with a reduced rate seen at higher glucose levels. Evaluation of the platform in vivo using a type 1 diabetic mouse model demonstrates efficacy in protecting against insulin-induced hypoglycemia by restoring blood glucose levels following an insulin overdose. The ability to tailor glucagon release in response to fluctuating glucose concentrations underscores the potential of this platform for improving glycemic control. These findings suggest that glucose-stabilized supramolecular peptide hydrogels hold significant promise for responsive drug delivery applications, offering an approach to manage glucose levels in diabetes and other metabolic disorders.more » « less
-
Abstract The management of diabetes in a manner offering autonomous insulin therapy responsive to glucose‐directed need, and moreover with a dosing schedule amenable to facile administration, remains an ongoing goal to improve the standard of care. While basal insulins with reduced dosing frequency, even once‐weekly administration, are on the horizon, there is still no approved therapy that offers glucose‐responsive insulin function. Herein, a nanoscale complex combining both electrostatic‐ and dynamic‐covalent interactions between a synthetic dendrimer carrier and an insulin analogue modified with a high‐affinity glucose‐binding motif yields an injectable insulin depot affording both glucose‐directed and long‐lasting insulin availability. Following a single injection, it is even possible to control blood glucose for at least one week in diabetic swine subjected to daily oral glucose challenges. Measurements of serum insulin concentration in response to challenge show increases in insulin corresponding to elevated blood glucose levels, an uncommon finding even in preclinical work on glucose‐responsive insulin. Accordingly, the subcutaneous nanocomplex that results from combining electrostatic‐ and dynamic‐covalent interactions between a modified insulin and a synthetic dendrimer carrier affords a glucose‐responsive insulin depot for week‐long control following a single routine injection.more » « less
-
Abstract Type 1 and advanced type 2 diabetes treatment involves daily injections or continuous infusion of exogenous insulin aimed at regulating blood glucose levels in the normoglycemic range. However, current options for insulin therapy are limited by the risk of hypoglycemia and are associated with suboptimal glycemic control outcomes. Therefore, a range of glucose‐responsive components that can undergo changes in conformation or show alterations in intermolecular binding capability in response to glucose stimulation has been studied for ultimate integration into closed‐loop insulin delivery or “smart insulin” systems. Here, an overview of the evolution and recent progress in the development of molecular approaches for glucose‐responsive insulin delivery systems, a rapidly growing subfield of precision medicine, is presented. Three central glucose‐responsive moieties, including glucose oxidase, phenylboronic acid, and glucose‐binding molecules are examined in detail. Future opportunities and challenges regarding translation are also discussed.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
