Artificial neural networks for closed loop control of in silico and ad hoc type 1 diabetes

Recent advances in the precision control strategy of artificial pancreas

The scientific diagnosis and treatment of patients with diabetes require frequent blood glucose testing and insulin delivery to normoglycemia. Therefore, an artificial pancreas with a continuous blood glucose (BG) monitoring function is an urgent research target in the medical industry. The problem of closed-loop algorithmic control of the BG with a time delay is a key and difficult issue that needs to be overcome in the development of an artificial pancreas. Firstly, the composition, structure, and control characteristics of the artificial pancreas are introduced. Subsequently, the research progress of artificial pancreas control algorithms is reviewed, and the characteristics, advantages, and disadvantages of proportional-integral-differential control, model predictive control, and artificial intelligence control are compared and analyzed to determine whether they are suitable for the practical application of the artificial pancreas. Additionally, key advancements in areas such as blood glucose data monitoring, adaptive models, wearable devices, and fully automated artificial pancreas systems are also reviewed. Finally, this review highlights that meal prediction, control safety, integration, streamlining the optimization of control algorithms, constant temperature preservation of insulin, and dual-hormone artificial pancreas are issues that require further attention in the future.

Offline reinforcement learning for safer blood glucose control in people with type 1 diabetes

The widespread adoption of effective hybrid closed loop systems would represent an important milestone of care for people living with type 1 diabetes (T1D). These devices typically utilise simple control algorithms to select the optimal insulin dose for maintaining blood glucose levels within a healthy range. Online reinforcement learning (RL) has been utilised as a method for further enhancing glucose control in these devices. Previous approaches have been shown to reduce patient risk and improve time spent in the target range when compared to classical control algorithms, but are prone to instability in the learning process, often resulting in the selection of unsafe actions. This work presents an evaluation of offline RL for developing effective dosing policies without the need for potentially dangerous patient interaction during training. This paper examines the utility of BCQ, CQL and TD3-BC in managing the blood glucose of the 30 virtual patients available within the FDA-approved UVA/Padova glucose dynamics simulator. When trained on less than a tenth of the total training samples required by online RL to achieve stable performance, this work shows that offline RL can significantly increase time in the healthy blood glucose range from 61.6±0.3% to 65.3±0.5% when compared to the strongest state-of-art baseline (p<0.001). This is achieved without any associated increase in low blood glucose events. Offline RL is also shown to be able to correct for common and challenging control scenarios such as incorrect bolus dosing, irregular meal timings and compression errors. The code for this work is available at: https://github.com/hemerson1/offline-glucose.

Artificial intelligence in critically ill diabetic patients: current status and future prospects

Recent years have witnessed increasing numbers of artificial intelligence (AI) based applications and devices being tested and approved for medical care. Diabetes is arguably the most common chronic disorder worldwide and AI is now being used for making an early diagnosis, to predict and diagnose early complications, increase adherence to therapy, and even motivate patients to manage diabetes and maintain glycemic control. However, these AI applications have largely been tested in non-critically ill patients and aid in managing chronic problems. Intensive care units (ICUs) have a dynamic environment generating huge data, which AI can extract and organize simultaneously, thus analysing many variables for diagnostic and/or therapeutic purposes in order to predict outcomes of interest. Even non-diabetic ICU patients are at risk of developing hypo or hyperglycemia, complicating their ICU course and affecting outcomes. In addition, to maintain glycemic control frequent blood sampling and insulin dose adjustments are required, increasing nursing workload and chances of error. AI has the potential to improve glycemic control while reducing the nursing workload and errors. Continuous glucose monitoring (CGM) devices, which are Food and Drug Administration (FDA) approved for use in non-critically ill patients, are now being recommended for use in specific ICU populations with increased accuracy. AI based devices including artificial pancreas and CGM regulated insulin infusion system have shown promise as comprehensive glycemic control solutions in critically ill patients. Even though many of these AI applications have shown potential, these devices need to be tested in larger number of ICU patients, have wider availability, show favorable cost-benefit ratio and be amenable for easy integration into the existing healthcare systems, before they become acceptable to ICU physicians for routine use.

Enhancing Clinical Translation of Cancer Using Nanoinformatics

Application of drugs in high doses has been required due to the limitations of no specificity, short circulation half-lives, as well as low bioavailability and solubility. Higher toxicity is the result of high dosage administration of drug molecules that increase the side effects of the drugs. Recently, nanomedicine, that is the utilization of nanotechnology in healthcare with clinical applications, has made many advancements in the areas of cancer diagnosis and therapy. To overcome the challenge of patient-specificity as well as time- and dose-dependency of drug administration, artificial intelligence (AI) can be significantly beneficial for optimization of nanomedicine and combinatorial nanotherapy. AI has become a tool for researchers to manage complicated and big data, ranging from achieving complementary results to routine statistical analyses. AI enhances the prediction precision of treatment impact in cancer patients and specify estimation outcomes. Application of AI in nanotechnology leads to a new field of study, i.e., nanoinformatics. Besides, AI can be coupled with nanorobots, as an emerging technology, to develop targeted drug delivery systems. Furthermore, by the advancements in the nanomedicine field, AI-based combination therapy can facilitate the understanding of diagnosis and therapy of the cancer patients. The main objectives of this review are to discuss the current developments, possibilities, and future visions in naoinformatics, for providing more effective treatment for cancer patients.

Blood glucose concentration prediction based on kernel canonical correlation analysis with particle swarm optimization and error compensation

BACKGROUND AND OBJECTIVE

Blood glucose levels in humans change over time. Continuous glucose monitoring system (CGMS), can constantly monitor the change of blood glucose concentration. Given the historical data of blood glucose, predicting the trend of blood glucose in a short term is important for diabetes. Appropriate behaviors can be adopted to prevent hypoglycemia or hyperglycemia.

METHODS

The method proposed in this paper only uses historical blood glucose data as input, rather than complex multi-dimensional input. Previous articles have demonstrated that canonical correlation analysis (CCA) can effectively predict blood glucose. The linear relationship between historical blood glucose values and predicted values was only considered regrettably. To compensate for this, this paper adds a kernel function to find out the non-linear relationship between blood glucose. In the introduced kernel function, some parameters need to be adjusted. To reduce the deviation caused by manual parameter adjustment, this paper discusses the role of particle swarm optimization (PSO). Besides, this article puts forward an error compensation for CCA to enhance the precision. Finally based on the prediction results of PSO-KCCA, a personalized hypoglycemic warning threshold is proposed.

RESULTS

The proposed method is validated using clinical data by the root mean square error (RMSE) and differential coefficient (R²). The average RMSE result in PSO-KCCA was 8.01, 11.98, 12.45, 13.23, 14.53, 16.40 mg/dL in prediction horizon (PH) =5, 10, 15, 20, 25, 30 min. The average R² was 0.95, 0.95, 0.98, 0.97, 0.98, and 0.97, respectively. The CCA with error compensation (EC-CCA) reduced RMSE by 33.45% compared with CCA. For the hypoglycemic warning, the average sensitivity obtained at 6 different PH values was 94.37%, and the specificity was 92.25%.

CONCLUSIONS

The experimental results confirm the effectiveness of PSO-KCCA in blood glucose prediction. The proposed EC-CCA successfully reduces the delay in the time series prediction. The personalized hypoglycemic warning threshold consider the influence of the model accuracy on the prediction results. This method guarantees the rate of underreporting during monitoring and ensures patient safety.

The significance of artificial intelligence in drug delivery system design

Over the last decade, increasing interest has been attracted towards the application of artificial intelligence (AI) technology for analyzing and interpreting the biological or genetic information, accelerated drug discovery, and identification of the selective small-molecule modulators or rare molecules and prediction of their behavior. Application of the automated workflows and databases for rapid analysis of the huge amounts of data and artificial neural networks (ANNs) for development of the novel hypotheses and treatment strategies, prediction of disease progression, and evaluation of the pharmacological profiles of drug candidates may significantly improve treatment outcomes. Target fishing (TF) by rapid prediction or identification of the biological targets might be of great help for linking targets to the novel compounds. AI and TF methods in association with human expertise may indeed revolutionize the current theranostic strategies, meanwhile, validation approaches are necessary to overcome the potential challenges and ensure higher accuracy. In this review, the significance of AI and TF in the development of drugs and delivery systems and the potential challenging issues have been highlighted.

Black-box Model Identification of Physical Activity in Type-l Diabetes Patients

In this paper we consider the problem of predicting future values of glucose in type-1 diabetes. In particular, we investigate the benefit of including physical activity, measured by an off-the-shelf wearable device, to other physiologic signals frequently used to predict blood-glucose concentration, namely injected insulin, carbohydrates intake, and past glucose samples measured by a Continuous Glucose Monitoring (CGM) sensor. Derivation of individualized predictors is crucial to cope with the wide inter- and intra-subject variability: learning and updating patient-specific models of the glucose-insulin system and using them to design personalized control actions has the potential to improve substantially patients' quality oflife. On data collected by 6 subjects for 5 days, we identify a black-box liner model that uses insulin and meal as inputs and glucose as output. Prediction Error Method (PEM) is used for parameter estimation. The personalized model is employed to derive patient-tailored predictors. This procedure is then repeated using a further physiological input, accounting for physical activity. The prediction accuracy of the two models, including or not physical activity, was compared on the basis of two metrics commonly used in system identification, namely Coefficient of Determination (COD) and Root Mean Squared Error. The models identified with physical activity have better performance, increasing the 3-hr prediction COD by mean ± standard deviation of 18.5% ± 30.1%.

Long-Term Glucose Forecasting Using a Physiological Model and Deconvolution of the Continuous Glucose Monitoring Signal

(1) Objective: Blood glucose forecasting in type 1 diabetes (T1D) management is a maturing field with numerous algorithms being published and a few of them having reached the commercialisation stage. However, accurate long-term glucose predictions (e.g., >60 min), which are usually needed in applications such as precision insulin dosing (e.g., an artificial pancreas), still remain a challenge. In this paper, we present a novel glucose forecasting algorithm that is well-suited for long-term prediction horizons. The proposed algorithm is currently being used as the core component of a modular safety system for an insulin dose recommender developed within the EU-funded PEPPER (Patient Empowerment through Predictive PERsonalised decision support) project. (2) Methods: The proposed blood glucose forecasting algorithm is based on a compartmental composite model of glucose-insulin dynamics, which uses a deconvolution technique applied to the continuous glucose monitoring (CGM) signal for state estimation. In addition to commonly employed inputs by glucose forecasting methods (i.e., CGM data, insulin, carbohydrates), the proposed algorithm allows the optional input of meal absorption information to enhance prediction accuracy. Clinical data corresponding to 10 adult subjects with T1D were used for evaluation purposes. In addition, in silico data obtained with a modified version of the UVa-Padova simulator was used to further evaluate the impact of accounting for meal absorption information on prediction accuracy. Finally, a comparison with two well-established glucose forecasting algorithms, the autoregressive exogenous (ARX) model and the latent variable-based statistical (LVX) model, was carried out. (3) Results: For prediction horizons beyond 60 min, the performance of the proposed physiological model-based (PM) algorithm is superior to that of the LVX and ARX algorithms. When comparing the performance of PM against the secondly ranked method (ARX) on a 120 min prediction horizon, the percentage improvement on prediction accuracy measured with the root mean square error, A-region of error grid analysis (EGA), and hypoglycaemia prediction calculated by the Matthews correlation coefficient, was 18.8 % , 17.9 % , and 80.9 % , respectively. Although showing a trend towards improvement, the addition of meal absorption information did not provide clinically significant improvements. (4) Conclusion: The proposed glucose forecasting algorithm is potentially well-suited for T1D management applications which require long-term glucose predictions.

Neural network-based model predictive control for type 1 diabetic rats on artificial pancreas system

Artificial pancreas system (APS) is a viable option to treat diabetic patients. Researchers, however, have not conclusively determined the best control method for APS. Due to intra-/inter-variability of insulin absorption and action, an individualized algorithm is required to control blood glucose level (BGL) for each patient. To this end, we developed model predictive control (MPC) based on artificial neural networks (ANNs), which combines ANN for BGL prediction based on inputs and MPC for BGL control based on the ANN (NN-MPC). First, we developed a mathematical model for diabetic rats, which was used to identify individual virtual subjects by fitting to empirical data collected through an APS, including BGL data, insulin injection, and food intake. Then, the virtual subjects were used to generate datasets for training ANNs. The NN-MPC determines control actions (insulin injection) based on BGL predicted by the ANN. To evaluate the NN-MPC, we conducted experiments using four virtual subjects under three different scenarios. Overall, the NN-MPC maintained BGL within the normal range about 90% of the time with a mean absolute deviation of 4.7 mg/dl from a desired BGL. Our findings suggest that the NN-MPC can provide subject-specific BGL control in conjunction with a closed-loop APS. Graphical abstract ᅟ.

Artificial Intelligence for Diabetes Management and Decision Support: Literature Review

Background

Artificial intelligence methods in combination with the latest technologies, including medical devices, mobile computing, and sensor technologies, have the potential to enable the creation and delivery of better management services to deal with chronic diseases. One of the most lethal and prevalent chronic diseases is diabetes mellitus, which is characterized by dysfunction of glucose homeostasis.

Objective

The objective of this paper is to review recent efforts to use artificial intelligence techniques to assist in the management of diabetes, along with the associated challenges.

Methods

A review of the literature was conducted using PubMed and related bibliographic resources. Analyses of the literature from 2010 to 2018 yielded 1849 pertinent articles, of which we selected 141 for detailed review.

Results

We propose a functional taxonomy for diabetes management and artificial intelligence. Additionally, a detailed analysis of each subject category was performed using related key outcomes. This approach revealed that the experiments and studies reviewed yielded encouraging results.

Conclusions

We obtained evidence of an acceleration of research activity aimed at developing artificial intelligence-powered tools for prediction and prevention of complications associated with diabetes. Our results indicate that artificial intelligence methods are being progressively established as suitable for use in clinical daily practice, as well as for the self-management of diabetes. Consequently, these methods provide powerful tools for improving patients’ quality of life.

An ARIMA Model With Adaptive Orders for Predicting Blood Glucose Concentrations and Hypoglycemia

The continuous glucose monitoring system is an effective tool, which enables the users to monitor their blood glucose (BG) levels. Based on the continuous glucose monitoring (CGM) data, we aim at predicting future BG levels so that appropriate actions can be taken in advance to prevent hyperglycemia or hypoglycemia. Due to the time-varying nonstationarity of CGM data, verified by Augmented Dickey-Fuller test and analysis of variance, an autoregressive integrated moving average (ARIMA) model with an adaptive identification algorithm of model orders is proposed in the prediction framework. Such identification algorithm adaptively determines the model orders and simultaneously estimates the corresponding parameters using Akaike Information Criterion and least square estimation. A case study is conducted with the CGM data of diabetics under daily living conditions to analyze the prediction performance of the proposed model together with the early hypoglycemic alarms. Results show that the proposed model outperforms the adaptive univariate model and ARIMA model.

Diabetes on demand and novel technologies

To date, the use of technology for the management of diabetes represents a promising area of innovation that can dramatically change diabetics' lives. In the past decade, the use of diabetes devices has widely grown and looks to have partially improved diabetes management. The combination of cloud technology with real-expert intervention saves time and improves efficiency, as well as empowering the patient. The application of mathematical models applied to diabetes therapy could lead to significant improvement in life quality and challenge the burden of hypoglycaemia. Events where an individual needs support are instantly achieved, triggering outreach alerts via cloud and wireless connectivity, thereby improving patient compliance and reducing disease costs.

Personalized blood glucose prediction: A hybrid approach using grammatical evolution and physiological models

The large patient variability in human physiology and the effects of variables such as exercise or meals challenge current prediction modeling techniques. Physiological models are very precise but they are typically complex and specific physiological knowledge is required. In contrast, data-based models allow the incorporation of additional inputs and accurately capture the relationship between these inputs and the outcome, but at the cost of losing the physiological meaning of the model. In this work, we designed a hybrid approach comprising physiological models for insulin and grammatical evolution, taking into account the clinical harm caused by deviations from the target blood glucose by using a penalizing fitness function based on the Clarke error grid. The prediction models were built using data obtained over 14 days for 100 virtual patients generated by the UVA/Padova T1D simulator. Midterm blood glucose was predicted for the 100 virtual patients using personalized models and different scenarios. The results obtained were promising; an average of 98.31% of the predictions fell in zones A and B of the Clarke error grid. Midterm predictions using personalized models are feasible when the configuration of grammatical evolution explored in this study is used. The study of new alternative models is important to move forward in the development of alarm-and-control applications for the management of type 1 diabetes and the customization of the patient's treatments. The hybrid approach can be adapted to predict short-term blood glucose values to detect continuous glucose-monitoring sensor errors and to estimate blood glucose values when the continuous glucose-monitoring system fails to provide them.

Chaotic time series prediction for glucose dynamics in type 1 diabetes mellitus using regime-switching models

In patients with type 1 diabetes mellitus (T1DM), glucose dynamics are influenced by insulin reactions, diet, lifestyle, etc., and characterized by instability and nonlinearity. With the objective of a dependable decision support system for T1DM self-management, we aim to model glucose dynamics using their nonlinear chaotic properties. A group of patients was monitored via continuous glucose monitoring (CGM) sensors for several days under free-living conditions. We assessed the glycemic variability (GV) and chaotic properties of each time series. Time series were subsequently transformed into the phase-space and individual autoregressive (AR) models were applied to predict glucose values over 30-minute and 60-minute prediction horizons (PH). The logistic smooth transition AR (LSTAR) model provided the best prediction accuracy for patients with high GV. For a PH of 30 minutes, the average values of root mean squared error (RMSE) and mean absolute error (MAE) for the LSTAR model in the case of patients in the hypoglycemia range were 5.83 ( ± 1.95) mg/dL and 5.18 ( ± 1.64) mg/dL, respectively. For a PH of 60 minutes, the average values of RMSE and MAE were 7.43 ( ± 1.87) mg/dL and 6.54 ( ± 1.6) mg/dL, respectively. Without the burden of measuring exogenous information, nonlinear regime-switching AR models provided fast and accurate results for glucose prediction.

A review of personalized blood glucose prediction strategies for T1DM patients

This paper presents a methodological review of models for predicting blood glucose (BG) concentration, risks and BG events. The surveyed models are classified into three categories, and they are presented in summary tables containing the most relevant data regarding the experimental setup for fitting and testing each model as well as the input signals and the performance metrics. Each category exhibits trends that are presented and discussed. This document aims to be a compact guide to determine the modeling options that are currently being exploited for personalized BG prediction.

Clinical Decision Support Tools: The Evolution of a Revolution

Dashboard systems for clinical decision support integrate data from multiple sources. These systems, the newest in a long line of dose calculators and other decision support tools, utilize Bayesian approaches to fully individualize dosing using information gathered through therapeutic drug monitoring. In the treatment of inflammatory bowel disease patients with infliximab, dashboards may reduce therapeutic failures and treatment costs. The history and future development of modern Bayesian dashboard systems is described.

Artificial Neural Networks and risk stratification models in Emergency Departments: The policy maker's perspective

The primary goal of Emergency Department (ED) physicians is to discriminate between individuals at low risk, who can be safely discharged, and patients at high risk, who require prompt hospitalization. The problem of correctly classifying patients is an issue involving not only clinical but also managerial aspects, since reducing the rate of admission of patients to EDs could dramatically cut costs. Nevertheless, a trade-off might arise due to the need to find a balance between economic interests and the health conditions of patients. This work considers patients in EDs after a syncope event and presents a comparative analysis between two models: a multivariate logistic regression model, as proposed by the scientific community to stratify the expected risk of severe outcomes in the short and long run, and Artificial Neural Networks (ANNs), an innovative model. The analysis highlights differences in correct classification of severe outcomes at 10 days (98.30% vs. 94.07%) and 1 year (97.67% vs. 96.40%), pointing to the superiority of Neural Networks. According to the results, there is also a significant superiority of ANNs in terms of false negatives both at 10 days (3.70% vs. 5.93%) and at 1 year (2.33% vs. 10.07%). However, considering the false positives, the adoption of ANNs would cause an increase in hospital costs, highlighting the potential trade-off which policy makers might face.

A therapy parameter-based model for predicting blood glucose concentrations in patients with type 1 diabetes

In this paper, the problem of predicting blood glucose concentrations (BG) for the treatment of patients with type 1 diabetes, is addressed. Predicting BG is of very high importance as most treatments, which consist in exogenous insulin injections, rely on the availability of BG predictions. Many models that can be used for predicting BG are available in the literature. However, it is widely admitted that it is almost impossible to perfectly model blood glucose dynamics while still being able to identify model parameters using only blood glucose measurements. The main contribution of this work is to propose a simple and identifiable linear dynamical model, which is based on the static prediction model of standard therapy. It is shown that the model parameters are intrinsically correlated with physician-set therapy parameters and that the reduction of the number of model parameters to identify leads to inferior data fits but to equivalent or slightly improved prediction capabilities compared to state-of-the-art models: a sign of an appropriate model structure and superior reliability. The validation of the proposed dynamic model is performed using data from the UVa simulator and real clinical data, and potential uses of the proposed model for state estimation and BG control are discussed.

Automated control of an adaptive bihormonal, dual-sensor artificial pancreas and evaluation during inpatient studies

Automated control of blood glucose in patients with type-1 diabetes has not yet been fully implemented. The aim of this study was to design and clinically evaluate a system that integrates a control algorithm with off-the-shelf subcutaneous sensors and pumps to automate the delivery of the hormones glucagon and insulin in response to continuous glucose sensor measurements. The automated component of the system runs an adaptive proportional derivative control algorithm which determines hormone delivery rates based on the sensed glucose measurements and the meal announcements by the patient. We provide details about the system design and the control algorithm, which incorporates both a fading memory proportional derivative controller (FMPD) and an adaptive system for estimating changing sensitivity to insulin based on a glucoregulatory model of insulin action. For an inpatient study carried out in eight subjects using Dexcom SEVEN PLUS sensors, prestudy HbA1c averaged 7.6, which translates to an estimated average glucose of 171 mg/dL. In contrast, during use of the automated system, after initial stabilization, glucose averaged 145 mg/dL and subjects were kept within the euglycemic range (between 70 and 180 mg/dL) for 73.1% of the time, indicating improved glycemic control. A further study on five additional subjects in which we used a newer and more reliable glucose sensor (Dexcom G4 PLATINUM) and made improvements to the insulin and glucagon pump communication system resulted in elimination of hypoglycemic events. For this G4 study, the system was able to maintain subjects' glucose levels within the near-euglycemic range for 71.6% of the study duration and the mean venous glucose level was 151 mg/dL.

Classification of intelligence quotient via brainwave sub-band power ratio features and artificial neural network

This paper elaborates on the novel intelligence assessment method using the brainwave sub-band power ratio features. The study focuses only on the left hemisphere brainwave in its relaxed state. Distinct intelligence quotient groups have been established earlier from the score of the Raven Progressive Matrices. Sub-band power ratios are calculated from energy spectral density of theta, alpha and beta frequency bands. Synthetic data have been generated to increase dataset from 50 to 120. The features are used as input to the artificial neural network. Subsequently, the brain behaviour model has been developed using an artificial neural network that is trained with optimized learning rate, momentum constant and hidden nodes. Findings indicate that the distinct intelligence quotient groups can be classified from the brainwave sub-band power ratios with 100% training and 88.89% testing accuracies.

MCF: a tool to find multi-scale community profiles in biological networks

Recent developments of complex graph clustering methods have implicated the practical applications with biological networks in different settings. Multi-scale Community Finder (MCF) is a tool to profile network communities (i.e., clusters of nodes) with the control of community sizes. The controlling parameter is referred to as the scale of the network community profile. MCF is able to find communities in all major types of networks including directed, signed, bipartite, and multi-slice networks. The fast computation promotes the practicability of the tool for large-scaled analysis (e.g., protein-protein interaction and gene co-expression networks). MCF is distributed as an open-source C++ package for academic use with both command line and user interface options, and can be downloaded at http://bsdxd.cpsc.ucalgary.ca/MCF. Detailed user manual and sample data sets are also available at the project website.