A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study

Minimal and Maximal Models to Quantitate Glucose Metabolism: Tools to Measure, to Simulate and to Run in Silico Clinical Trials

Several models have been proposed to describe the glucose system at whole-body, organ/tissue and cellular level, designed to measure non-accessible parameters (minimal models), to simulate system behavior and run in silico clinical trials (maximal models). Here, we will review the authors' work, by putting it into a concise historical background. We will discuss first the parametric portrait provided by the oral minimal models-building on the classical intravenous glucose tolerance test minimal models-to measure otherwise non-accessible key parameters like insulin sensitivity and beta-cell responsivity from a physiological oral test, the mixed meal or the oral glucose tolerance tests, and what can be gained by adding a tracer to the oral glucose dose. These models were used in various pathophysiological studies, which we will briefly review. A deeper understanding of insulin sensitivity can be gained by measuring insulin action in the skeletal muscle. This requires the use of isotopic tracers: both the classical multiple-tracer dilution and the positron emission tomography techniques are discussed, which quantitate the effect of insulin on the individual steps of glucose metabolism, that is, bidirectional transport plasma-interstitium, and phosphorylation. Finally, we will present a cellular model of insulin secretion that, using a multiscale modeling approach, highlights the relations between minimal model indices and subcellular secretory events. In terms of maximal models, we will move from a parametric to a flux portrait of the system by discussing the triple tracer meal protocol implemented with the tracer-to-tracee clamp technique. This allows to arrive at quasi-model independent measurement of glucose rate of appearance (Ra), endogenous glucose production (EGP), and glucose rate of disappearance (Rd). Both the fast absorbing simple carbs and the slow absorbing complex carbs are discussed. This rich data base has allowed us to build the UVA/Padova Type 1 diabetes and the Padova Type 2 diabetes large scale simulators. In particular, the UVA/Padova Type 1 simulator proved to be a very useful tool to safely and effectively test in silico closed-loop control algorithms for an artificial pancreas (AP). This was the first and unique simulator of the glucose system accepted by the U.S. Food and Drug Administration as a substitute to animal trials for in silico testing AP algorithms. Recent uses of the simulator have looked at glucose sensors for non-adjunctive use and new insulin molecules.

Personalized insulin dose manipulation attack and its detection using interval-based temporal patterns and machine learning algorithms

Many patients with diabetes are currently being treated with insulin pumps and other diabetes devices which improve their quality of life and enable effective treatment of diabetes. These devices are connected wirelessly and thus, are vulnerable to cyber-attacks which have already been proven feasible. In this paper, we focus on two types of cyber-attacks on insulin pump systems: an overdose of insulin, which can cause hypoglycemia, and an underdose of insulin, which can cause hyperglycemia. Both of these attacks can result in a variety of complications and endanger a patient's life. Specifically, we propose a sophisticated and personalized insulin dose manipulation attack; this attack is based on a novel method of predicting the blood glucose (BG) level in response to insulin dose administration. To protect patients from the proposed sophisticated and malicious insulin dose manipulation attacks, we also present an automated machine learning based system for attack detection; the detection system is based on an advanced temporal pattern mining process, which is performed on the logs of real insulin pumps and continuous glucose monitors (CGMs). Our multivariate time-series data (MTSD) collection consists of 225,780 clinical logs, collected from real insulin pumps and CGMs of 47 patients with type I diabetes (13 adults and 34 children) from two different clinics at Soroka University Medical Center in Beer-Sheva, Israel over a four-year period. We enriched our data collection with additional relevant medical information related to the subjects. In the extensive experiments performed, we evaluated the proposed attack and detection system and examined whether: (1) it is possible to accurately predict BG levels in order to create malicious data that simulate a manipulation attack and the patient's body in response to it; (2) it is possible to automatically detect such attacks based on advanced machine learning (ML) methods that leverage temporal patterns; (3) the detection capabilities of the proposed detection system differ for insulin overdose and underdose attacks; and (4) the granularity of the learning model (general / adult vs. pediatric clinic / individual patient) affects the detection capabilities. Our results show that (a) it is possible to predict, with nearly 90% accuracy, BG levels using our proposed methods, and by doing so, enable malicious data creation for our detection system evaluation; (b) it is possible to accurately detect insulin manipulation attacks using temporal patterns mining using several ML methods, including Logistic Regression, Random Forest, TPF class model, TPF top k, and ANN algorithms; (c) it is easier to detect an overdose attack than an underdose attack in more than 25%, in terms of AUC scores; and (d) the adult vs. pediatric model outperformed models of other granularities in the detection of overdose attacks, while the general model outperformed the other models in the case of detecting underdose attacks; for both attacks, attack detection among children was found to be more challenging than among adults. In addition to its use in the evaluation of our detection system, the proposed BG prediction method has great importance in the medical domain where it can contribute to improved care of patients with diabetes.

The challenge of predicting blood glucose concentration changes in patients with type I diabetes

Patients with Type I Diabetes (T1D) must take insulin injections to prevent the serious long term effects of hyperglycemia. They must also be careful not to inject too much insulin because this could induce (potentially fatal) hypoglycemia. Patients therefore follow a "regimen" that determines how much insulin to inject at each time, based on various measurements. We can produce an effective regimen if we can accurately predict a patient's future blood glucose (BG) values from his/her current features. This study explores the challenges of predicting future BG by applying a number of machine learning algorithms, as well as various data preprocessing variations (corresponding to 312 [learner, preprocessed-dataset] combinations), to a new T1D dataset that contains 29,601 entries from 47 different patients. Our most accurate predictor, a weighted ensemble of two Gaussian Process Regression models, achieved a (cross-validation) errL1 loss of 2.7 mmol/L (48.65 mg/dl). This result was unexpectedly poor given that one can obtain an errL1 of 2.9 mmol/L (52.43 mg/dl) using the naive approach of simply predicting the patient's average BG. These results suggest that the diabetes diary data that is typically collected may be insufficient to produce accurate BG prediction models; additional data may be necessary to build accurate BG prediction models over hours.

Artificial Intelligence in Decision Support Systems for Type 1 Diabetes

Type 1 diabetes (T1D) is a chronic health condition resulting from pancreatic beta cell dysfunction and insulin depletion. While automated insulin delivery systems are now available, many people choose to manage insulin delivery manually through insulin pumps or through multiple daily injections. Frequent insulin titrations are needed to adequately manage glucose, however, provider adjustments are typically made every several months. Recent automated decision support systems incorporate artificial intelligence algorithms to deliver personalized recommendations regarding insulin doses and daily behaviors. This paper presents a comprehensive review of computational and artificial intelligence-based decision support systems to manage T1D. Articles were obtained from PubMed, IEEE Xplore, and ScienceDirect databases. No time period restrictions were imposed on the search. After removing off-topic articles and duplicates, 562 articles were left to review. Of those articles, we identified 61 articles for comprehensive review based on algorithm evaluation using real-world human data, in silico trials, or clinical studies. We grouped decision support systems into general categories of (1) those which recommend adjustments to insulin and (2) those which predict and help avoid hypoglycemia. We review the artificial intelligence methods used for each type of decision support system, and discuss the performance and potential applications of these systems.

Utility of Big Data in Predicting Short-Term Blood Glucose Levels in Type 1 Diabetes Mellitus Through Machine Learning Techniques

Machine learning techniques combined with wearable electronics can deliver accurate short-term blood glucose level prediction models. These models can learn personalized glucose–insulin dynamics based on the sensor data collected by monitoring several aspects of the physiological condition and daily activity of an individual. Until now, the prevalent approach for developing data-driven prediction models was to collect as much data as possible to help physicians and patients optimally adjust therapy. The objective of this work was to investigate the minimum data variety, volume, and velocity required to create accurate person-centric short-term prediction models. We developed a series of these models using different machine learning time series forecasting techniques suitable for execution within a wearable processor. We conducted an extensive passive patient monitoring study in real-world conditions to build an appropriate data set. The study involved a subset of type 1 diabetic subjects wearing a flash glucose monitoring system. We comparatively and quantitatively evaluated the performance of the developed data-driven prediction models and the corresponding machine learning techniques. Our results indicate that very accurate short-term prediction can be achieved by only monitoring interstitial glucose data over a very short time period and using a low sampling frequency. The models developed can predict glucose levels within a 15-min horizon with an average error as low as 15.43 mg/dL using only 24 historic values collected within a period of sex hours, and by increasing the sampling frequency to include 72 values, the average error is reduced to 10.15 mg/dL. Our prediction models are suitable for execution within a wearable device, requiring the minimum hardware requirements while at simultaneously achieving very high prediction accuracy.

Using LSTMs to learn physiological models of blood glucose behavior

For people with type 1 diabetes, good blood glucose control is essential to keeping serious disease complications at bay. This entails carefully monitoring blood glucose levels and taking corrective steps whenever they are too high or too low. If blood glucose levels could be accurately predicted, patients could take proactive steps to prevent blood glucose excursions from occurring. However, accurate predictions require complex physiological models of blood glucose behavior. Factors such as insulin boluses, carbohydrate intake, and exercise influence blood glucose in ways that are difficult to capture through manually engineered equations. In this paper, we describe a recursive neural network (RNN) approach that uses long short-term memory (LSTM) units to learn a physiological model of blood glucose. When trained on raw data from real patients, the LSTM networks (LSTMs) obtain results that are competitive with a previous state-of-the-art model based on manually engineered physiological equations. The RNN approach can incorporate arbitrary physiological parameters without the need for sophisticated manual engineering, thus holding the promise of further improvements in prediction accuracy.

Next-generation, personalised, model-based critical care medicine: a state-of-the art review of in silico virtual patient models, methods, and cohorts, and how to validation them

Critical care, like many healthcare areas, is under a dual assault from significantly increasing demographic and economic pressures. Intensive care unit (ICU) patients are highly variable in response to treatment, and increasingly aging populations mean ICUs are under increasing demand and their cohorts are increasingly ill. Equally, patient expectations are growing, while the economic ability to deliver care to all is declining. Better, more productive care is thus the big challenge. One means to that end is personalised care designed to manage the significant inter- and intra-patient variability that makes the ICU patient difficult. Thus, moving from current "one size fits all" protocolised care to adaptive, model-based "one method fits all" personalised care could deliver the required step change in the quality, and simultaneously the productivity and cost, of care. Computer models of human physiology are a unique tool to personalise care, as they can couple clinical data with mathematical methods to create subject-specific models and virtual patients to design new, personalised and more optimal protocols, as well as to guide care in real-time. They rely on identifying time varying patient-specific parameters in the model that capture inter- and intra-patient variability, the difference between patients and the evolution of patient condition. Properly validated, virtual patients represent the real patients, and can be used in silico to test different protocols or interventions, or in real-time to guide care. Hence, the underlying models and methods create the foundation for next generation care, as well as a tool for safely and rapidly developing personalised treatment protocols over large virtual cohorts using virtual trials. This review examines the models and methods used to create virtual patients. Specifically, it presents the models types and structures used and the data required. It then covers how to validate the resulting virtual patients and trials, and how these virtual trials can help design and optimise clinical trial. Links between these models and higher order, more complex physiome models are also discussed. In each section, it explores the progress reported up to date, especially on core ICU therapies in glycemic, circulatory and mechanical ventilation management, where high cost and frequency of occurrence provide a significant opportunity for model-based methods to have measurable clinical and economic impact. The outcomes are readily generalised to other areas of medical care.

In silico assessment of biomedical products: The conundrum of rare but not so rare events in two case studies

In silico clinical trials, defined as "The use of individualized computer simulation in the development or regulatory evaluation of a medicinal product, medical device, or medical intervention," have been proposed as a possible strategy to reduce the regulatory costs of innovation and the time to market for biomedical products. We review some of the the literature on this topic, focusing in particular on those applications where the current practice is recognized as inadequate, as for example, the detection of unexpected severe adverse events too rare to be detected in a clinical trial, but still likely enough to be of concern. We then describe with more details two case studies, two successful applications of in silico clinical trial approaches, one relative to the University of Virginia/Padova simulator that the Food and Drug Administration has accepted as possible replacement for animal testing in the preclinical assessment of artificial pancreas technologies, and the second, an investigation of the probability of cardiac lead fracture, where a Bayesian network was used to combine in vivo and in silico observations, suggesting a whole new strategy of in silico-augmented clinical trials, to be used to increase the numerosity where recruitment is impossible, or to explore patients' phenotypes that are unlikely to appear in the trial cohort, but are still frequent enough to be of concern.

Real-Time Statistical Modeling of Blood Sugar

Diabetes is considered a chronic disease that incurs various types of cost to the world. One major challenge in the control of Diabetes is the real time determination of the proper insulin dose. In this paper, we develop a prototype for real time blood sugar control, integrated with the cloud. Our system controls blood sugar by observing the blood sugar level and accordingly determining the appropriate insulin dose based on patient's historical data, all in real time and automatically. To determine the appropriate insulin dose, we propose two statistical models for modeling blood sugar profiles, namely ARIMA and Markov-based model. Our experiment used to evaluate the performance of the two models shows that the ARIMA model outperforms the Markov-based model in terms of prediction accuracy.

Advancing our understanding of the glucose system via modeling: a perspective

The glucose story begins with Claude Bernard's discovery of glycogen and milieu interieur, continued with Banting's and Best's discovery of insulin and with Rudolf Schoenheimer's paradigm of dynamic body constituents. Tracers and compartmental models allowed moving to the first quantitative pictures of the system and stimulated important developments in terms of modeling methodology. Three classes of multiscale models, models to measure, models to simulate, and models to control the glucose system, are reviewed in their historical development with an eye to the future.

Temporal abstraction and temporal Bayesian networks in clinical domains: a survey

OBJECTIVES

Temporal abstraction (TA) of clinical data aims to abstract and interpret clinical data into meaningful higher-level interval concepts. Abstracted concepts are used for diagnostic, prediction and therapy planning purposes. On the other hand, temporal Bayesian networks (TBNs) are temporal extensions of the known probabilistic graphical models, Bayesian networks. TBNs can represent temporal relationships between events and their state changes, or the evolution of a process, through time. This paper offers a survey on techniques/methods from these two areas that were used independently in many clinical domains (e.g. diabetes, hepatitis, cancer) for various clinical tasks (e.g. diagnosis, prognosis). A main objective of this survey, in addition to presenting the key aspects of TA and TBNs, is to point out important benefits from a potential integration of TA and TBNs in medical domains and tasks. The motivation for integrating these two areas is their complementary function: TA provides clinicians with high level views of data while TBNs serve as a knowledge representation and reasoning tool under uncertainty, which is inherent in all clinical tasks.

METHODS

Key publications from these two areas of relevance to clinical systems, mainly circumscribed to the latest two decades, are reviewed and classified. TA techniques are compared on the basis of: (a) knowledge acquisition and representation for deriving TA concepts and (b) methodology for deriving basic and complex temporal abstractions. TBNs are compared on the basis of: (a) representation of time, (b) knowledge representation and acquisition, (c) inference methods and the computational demands of the network, and (d) their applications in medicine.

RESULTS

The survey performs an extensive comparative analysis to illustrate the separate merits and limitations of various TA and TBN techniques used in clinical systems with the purpose of anticipating potential gains through an integration of the two techniques, thus leading to a unified methodology for clinical systems. The surveyed contributions are evaluated using frameworks of respective key features. In addition, for the evaluation of TBN methods, a unifying clinical domain (diabetes) is used.

CONCLUSION

The main conclusion transpiring from this review is that techniques/methods from these two areas, that so far are being largely used independently of each other in clinical domains, could be effectively integrated in the context of medical decision-support systems. The anticipated key benefits of the perceived integration are: (a) during problem solving, the reasoning can be directed at different levels of temporal and/or conceptual abstractions since the nodes of the TBNs can be complex entities, temporally and structurally and (b) during model building, knowledge generated in the form of basic and/or complex abstractions, can be deployed in a TBN.

Analysis of intravenous glucose tolerance test data using parametric and nonparametric modeling: application to a population at risk for diabetes

BACKGROUND

Modeling studies of the insulin-glucose relationship have mainly utilized parametric models, most notably the minimal model (MM) of glucose disappearance. This article presents results from the comparative analysis of the parametric MM and a nonparametric Laguerre based Volterra Model (LVM) applied to the analysis of insulin modified (IM) intravenous glucose tolerance test (IVGTT) data from a clinical study of gestational diabetes mellitus (GDM).

METHODS

An IM IVGTT study was performed 8 to 10 weeks postpartum in 125 women who were diagnosed with GDM during their pregnancy [population at risk of developing diabetes (PRD)] and in 39 control women with normal pregnancies (control subjects). The measured plasma glucose and insulin from the IM IVGTT in each group were analyzed via a population analysis approach to estimate the insulin sensitivity parameter of the parametric MM. In the nonparametric LVM analysis, the glucose and insulin data were used to calculate the first-order kernel, from which a diagnostic scalar index representing the integrated effect of insulin on glucose was derived.

RESULTS

Both the parametric MM and nonparametric LVM describe the glucose concentration data in each group with good fidelity, with an improved measured versus predicted r² value for the LVM of 0.99 versus 0.97 for the MM analysis in the PRD. However, application of the respective diagnostic indices of the two methods does result in a different classification of 20% of the individuals in the PRD.

CONCLUSIONS

It was found that the data based nonparametric LVM revealed additional insights about the manner in which infused insulin affects blood glucose concentration.

A fast generalizable solution method for glucose control algorithms

In critical care tight control of blood glucose levels has been shown to lead to better clinical outcomes. The need to develop new protocols for tight glucose control, as well as the opportunity to optimize a variety of other drug therapies, has led to resurgence in model-based medical decision support in this area. One still valid hindrance to developing new model-based protocols using so-called virtual patients, retrospective clinical data, and Monte Carlo methods is the large amount of computational time and resources needed. This paper develops fast analytical-based methods for insulin-glucose system model that are generalizable to other similar systems. Exploiting the structure and partial solutions in a subset of the model is the key in finding accurate fast solutions to the full model. This approach successfully reduced computing time by factors of 5600-144000 depending on the numerical error management method, for large (50-164 patients) virtual trials and Monte Carlo analysis. It thus allows new model-based or model-derived protocols to be rapidly developed via extensive simulation. The new method is rigorously compared to existing standard numerical solutions and is found to be highly accurate to within 0.2%.

Neural-network models for the blood glucose metabolism of a diabetic

We study the application of neural networks to modeling the blood glucose metabolism of a diabetic. In particular we consider recurrent neural networks and time series convolution neural networks which we compare to linear models and to nonlinear compartment models. We include a linear error model to take into account the uncertainty in the system and for handling missing blood glucose observations. Our results indicate that best performance can be achieved by the combination of the recurrent neural network and the linear error model.

The Glucosafe system for tight glycemic control in critical care: a pilot evaluation study

PURPOSE

"Glucosafe' is a new model-based decision support system for glycemic control in critical care. Safety and achievement of glycemic goals using the system are tested prospectively.

METHODS

Four penalty functions were developed to balance regimens of nutrition and insulin therapy against model-predicted glycemic outcome. The system advises the regimen where the penalty sum is minimal. An interactive interface allows advice alterations. Ten hyperglycemic patients (median Acute Physiology and Chronic Health Evaluation II, 12.5; interquartile range, 7.5-16.3) from a neuro and trauma intensive care unit were included for pilot testing using Glucosafe for 12 to 14 hours. Glycemic outcomes were compared to the 24-hour intervals before and after intervention.

RESULTS

Hypoglycemia (blood glucose [BG] <3.5 mmol/L) was not observed. Mean log-normal BG +/- standard deviation was reduced from 8.6 +/- 2.4 mmol/L preintervention to 7.0 +/- 1.1 mmol/L during the intervention. Nine patients reached the 4.4- to 6.1-mmol/L band after a mean 5 hours. At 5 hours intervention, mean log-normal BG was 6.7 mmol/L, 40% of measurements were in the 4.4- to 6.1-mmol/L band, and 84% were in the 4.4- to 7.75-mmol/L band.

CONCLUSIONS

Safety was demonstrated with the developed penalty functions. The low BG variance achieved may permit minor adjustments of the penalty function values to reduce average BG if desired.

Simulation environment to evaluate closed-loop insulin delivery systems in type 1 diabetes

BACKGROUND

Closed-loop insulin delivery systems linking subcutaneous insulin infusion to real-time continuous glucose monitoring need to be evaluated in humans, but progress can be accelerated with the use of in silico testing. We present a simulation environment designed to support the development and testing of closed-loop insulin delivery systems in type 1 diabetes mellitus (T1DM).

METHODS

The principal components of the simulation environment include a mathematical model of glucose regulation representing a virtual population with T1DM, the glucose measurement model, and the insulin delivery model. The simulation environment is highly flexible. The user can specify an experimental protocol, define a population of virtual subjects, choose glucose measurement and insulin delivery models, and specify outcome measures. The environment provides graphical as well as numerical outputs to enable a comprehensive analysis of in silico study results. The simulation environment is validated by comparing its predictions against a clinical study evaluating overnight closed-loop insulin delivery in young people with T1DM using a model predictive controller.

RESULTS

The simulation model of glucose regulation is described, and population values of 18 synthetic subjects are provided. The validation study demonstrated that the simulation environment was able to reproduce the population results of the clinical study conducted in young people with T1DM.

CONCLUSIONS

Closed-loop trials in humans should be preceded and concurrently guided by highly efficient and resource-saving computer-based simulations. We demonstrate validity of population-based predictions obtained with our simulation environment.

A Nonlinear State Space Model for the Blood Glucose Metabolism of a Diabetic (Ein nichtlineares Zustandsraummodell für den Blutglukosemetabolismus eines Diabetikers)

The blood glucose metabolism of a diabetic is a complex nonlinear process closely linked to a number of internal factors which are not easily accessible to measurement. Based on accessible information – such as occasional blood glucose measurements and information about food intake and physical exercise – the system appears highly stochastic and the quantity of main interest, the blood glucose concentration, is very difficult to model and to predict. In this paper we describe a stochastic nonlinear state space model for modeling the blood glucose concentration of a diabetic patient. The model structure is based on physiological prior knowledge and the main nonlinearities are modeled using artificial neural networks. Offline training of the model is performed using a newly developed Monte-Carlo generalized EM (expectation maximization) algorithm. Online prediction is performed using particle filters. Our experimental results show that our approach provides better prediction results than a number of competing approaches.

Nonlinear modeling of the dynamic effects of infused insulin on glucose: comparison of compartmental with Volterra models

This paper presents the results of a computational study that compares simulated compartmental (differential equation) and Volterra models of the dynamic effects of insulin on blood glucose concentration in humans. In the first approach, we employ the widely accepted "minimal model" and an augmented form of it, which incorporates the effect of insulin secretion by the pancreas, in order to represent the actual closed-loop operating conditions of the system, and in the second modeling approach, we employ the general class of Volterra-type models that are estimated from input-output data. We demonstrate both the equivalence between the two approaches analytically and the feasibility of obtaining accurate Volterra models from insulin-glucose data generated from the compartmental models. The results corroborate the proposition that it may be preferable to obtain data-driven (i.e., inductive) models in a more general and realistic operating context, without resorting to the restrictive prior assumptions and simplifications regarding model structure and/or experimental protocols (e.g., glucose tolerance tests) that are necessary for the compartmental models proposed previously. These prior assumptions may lead to results that are improperly constrained or biased by preconceived (and possibly erroneous) notions-a risk that is avoided when we let the data guide the inductive selection of the appropriate model within the general class of Volterra-type models, as our simulation results suggest.

Diabetes: Models, Signals, and Control

The control of diabetes is an interdisciplinary endeavor, which includes a significant biomedical engineering component, with traditions of success beginning in the early 1960s. It began with modeling of the insulin-glucose system, and progressed to large-scale in silico experiments, and automated closed-loop control (artificial pancreas). Here, we follow these engineering efforts through the last, almost 50 years. We begin with the now classic minimal modeling approach and discuss a number of subsequent models, which have recently resulted in the first in silico simulation model accepted as substitute to animal trials in the quest for optimal diabetes control. We then review metabolic monitoring, with a particular emphasis on the new continuous glucose sensors, on the analyses of their time-series signals, and on the opportunities that they present for automation of diabetes control. Finally, we review control strategies that have been successfully employed in vivo or in silico, presenting a promise for the development of a future artificial pancreas and, in particular, discuss a modular architecture for building closed-loop control systems, including insulin delivery and patient safety supervision layers. We conclude with a brief discussion of the unique interactions between human physiology, behavioral events, engineering modeling and control relevant to diabetes.

A subcutaneous insulin pharmacokinetic model for computer simulation in a diabetes decision support role: model structure and parameter identification

OBJECTIVE

The goal of this study was to develop a unified physiological subcutaneous (SC) insulin absorption model for computer simulation in a clinical diabetes decision support role. The model must model the plasma insulin appearance of a wide range of current insulins, especially monomer insulin and insulin glargine, utilizing common chemical states and transport rates, where appropriate.

METHODS

A compartmental model was developed with 13 patient-specific model parameters covering six diverse insulin types [rapid-acting, regular, neutral protamine Hagedorn (NPH), lente, ultralente, and glargine insulin]. Model parameters were identified using 37 sets of mean plasma insulin time-course data from an extensive literature review via nonlinear optimization methods.

RESULTS

All fitted parameters have a coefficient of variation <100% (median 51.3%, 95th percentile 3.6-60.6%) and can be considered a posteriori identifiable.

CONCLUSION

A model is presented to describe SC injected insulin appearance in plasma in a diabetes decision support role. Clinically current insulin types (monomeric insulin, regular insulin, NPH, insulin, and glargine) and older insulin types (lente and ultralente) are included in a unified framework that accounts for nonlinear concentration and dose dependency. Future work requires clinical validation using published pharmacokinetic studies.

The impact of parameter identification methods on drug therapy control in an intensive care unit

This paper investigates the impact of fast parameter identification methods, which do not require any forward simulations, on model-based glucose control, using retrospective data in the Christchurch Hospital Intensive Care Unit. The integral-based identification method has been previously clinically validated and extensively applied in a number of biomedical applications; and is a crucial element in the presented model-based therapeutics approach. Common non-linear regression and gradient descent approaches are too computationally intense and not suitable for the glucose control applications presented. The main focus in this paper is on better characterizing and understanding the importance of the integral in the formulation and the effect it has on model-based drug therapy control. As a comparison, a potentially more natural derivative formulation which has the same computation speed advantages is investigated, and is shown to go unstable with respect to modelling error which is always present clinically. The integral method remains robust.

Neural network based glucose - insulin metabolism models for children with Type 1 diabetes

In this paper two models for the simulation of glucose-insulin metabolism of children with Type 1 diabetes are presented. The models are based on the combined use of Compartmental Models (CMs) and artificial Neural Networks (NNs). Data from children with Type 1 diabetes, stored in a database, have been used as input to the models. The data are taken from four children with Type 1 diabetes and contain information about glucose levels taken from continuous glucose monitoring system, insulin intake and food intake, along with corresponding time. The influences of taken insulin on plasma insulin concentration, as well as the effect of food intake on glucose input into the blood from the gut, are estimated from the CMs. The outputs of CMs, along with previous glucose measurements, are fed to a NN, which provides short-term prediction of glucose values. For comparative reasons two different NN architectures have been tested: a Feed-Forward NN (FFNN) trained with the back-propagation algorithm with adaptive learning rate and momentum, and a Recurrent NN (RNN), trained with the Real Time Recurrent Learning (RTRL) algorithm. The results indicate that the best prediction performance can be achieved by the use of RNN.

Dynamic Bayesian networks as prognostic models for clinical patient management

Prognostic models in medicine are usually been built using simple decision rules, proportional hazards models, or Markov models. Dynamic Bayesian networks (DBNs) offer an approach that allows for the incorporation of the causal and temporal nature of medical domain knowledge as elicited from domain experts, thereby allowing for detailed prognostic predictions. The aim of this paper is to describe the considerations that must be taken into account when constructing a DBN for complex medical domains and to demonstrate their usefulness in practice. To this end, we focus on the construction of a DBN for prognosis of carcinoid patients, compare performance with that of a proportional hazards model, and describe predictions for three individual patients. We show that the DBN can make detailed predictions, about not only patient survival, but also other variables of interest, such as disease progression, the effect of treatment, and the development of complications. Strengths and limitations of our approach are discussed and compared with those offered by traditional methods.

Meal simulation model of the glucose-insulin system

A simulation model of the glucose-insulin system in the postprandial state can be useful in several circumstances, including testing of glucose sensors, insulin infusion algorithms and decision support systems for diabetes. Here, we present a new simulation model in normal humans that describes the physiological events that occur after a meal, by employing the quantitative knowledge that has become available in recent years. Model parameters were set to fit the mean data of a large normal subject database that underwent a triple tracer meal protocol which provided quasi-model-independent estimates of major glucose and insulin fluxes, e.g., meal rate of appearance, endogenous glucose production, utilization of glucose, insulin secretion. By decomposing the system into subsystems, we have developed parametric models of each subsystem by using a forcing function strategy. Model results are shown in describing both a single meal and normal daily life (breakfast, lunch, dinner) in normal. The same strategy is also applied on a smaller database for extending the model to type 2 diabetes.

GIM, simulation software of meal glucose-insulin model

BACKGROUND

A simulation model of the glucose-insulin system in normal life conditions can be very useful in diabetes research, e.g., testing insulin infusion algorithms and decision support systems and assessing glucose sensor performance and patient and student training. A new meal simulation model has been proposed that incorporates state-of-the-art quantitative knowledge on glucose metabolism and its control by insulin at both organ/tissue and whole-body levels. This article presents the interactive simulation software GIM (glucose insulin model), which implements this model.

METHODS

The model is implemented in MATLAB, version 7.0.1, and is designed with a windows interface that allows the user to easily simulate a 24-hour daily life of a normal, type 2, or type 1 diabetic subject. A Simulink version is also available. Three meals a day are considered. Both open- and closed-loop controls are available for simulating a type 1 diabetic subject.

RESULTS

Software options are described in detail. Case studies are presented to illustrate the potential of the software, e.g., compare a normal subject vs an insulin-resistant subject or open-loop vs closed-loop insulin infusion in type 1 diabetes treatment.

CONCLUSIONS

User-friendly software that implements a state-of-the-art physiological model of the glucose-insulin system during a meal has been presented. The GIM graphical interface makes its use extremely easy for investigators without specific expertise in modeling.

A real time simulation model of glucose-insulin metabolism for type 1 diabetes patients

In this paper, a simulation model of glucose-insulin metabolism for Type 1 diabetes patients is presented. The proposed system is based on the combination of Compartmental Models (CMs) and artificial Neural Networks (NNs). This model aims at the development of an accurate system, in order to assist Type 1 diabetes patients to handle their blood glucose profile and recognize dangerous metabolic states. Data from a Type 1 diabetes patient, stored in a database, have been used as input to the hybrid system. The data contain information about measured blood glucose levels, insulin intake, and description of food intake, along with the corresponding time. The data are passed to three separate CMs, which produce estimations about (i) the effect of Short Acting (SA) insulin intake on blood insulin concentration, (ii) the effect of Intermediate Acting (IA) insulin intake on blood insulin concentration, and (iii) the effect of carbohydrate intake on blood glucose absorption from the gut. The outputs of the three CMs are passed to a Recurrent NN (RNN) in order to predict subsequent blood glucose levels. The RNN is trained with the Real Time Recurrent Learning (RTRL) algorithm. The resulted blood glucose predictions are promising for the use of the proposed model for blood glucose level estimation for Type 1 diabetes patients.

A stochastic model to assess the variability of blood glucose time series in diabetic patients self-monitoring

Several studies have shown that patients suffering from Diabetes Mellitus can significantly delay the onset and slow down the progression of diabetes micro- and macro-angiopathic complications through intensive monitoring and treatment. In general, intensive treatments imply a careful blood glucose level (BGL) self-monitoring. The analysis of BGL measurements is one of the most important tasks in order to assess the glucose metabolic control and to revise the therapeutic protocol. Recent clinical studies have shown the correlation between the glucose variability and the long-term diabetes related complications. In this paper, we propose a stochastic model to extract the time course of such variability from the self-monitoring BGL time series. This information can be conveniently combined with other analysis to evaluate the adequacy of the therapeutic protocol and to highlight periods characterized by an increasing glucose instability. The method here proposed has been validated on two simulated data sets and tested with success in the retrospective analysis of three patients' data sets.

Neural network modeling and control of type 1 diabetes mellitus

This paper presents a developed and validated dynamic simulation model of type 1 diabetes, that simulates the progression of the disease and the two term controller that is responsible for the insulin released to stabilize the glucose level. The modeling and simulation of type 1 diabetes mellitus is based on an artificial neural network approach. The methodology builds upon an existing rich database on the progression of type 1 diabetes for a group of diabetic patients. The model was found to perform well at estimating the next glucose level over time without control. A neural controller that mimics the pancreas secretion of insulin into the body was also developed. This controller is of the two term type: one stage is responsible for short-term and the other for mid-term insulin delivery. It was found that the controller designed predicts an adequate amount of insulin that should be delivered into the body to obtain a normalization of the elevated glucose level. This helps to achieve the main objective of insulin therapy: to obtain an accurate estimate of the amount of insulin to be delivered in order to compensate for the increase in glucose concentration.

Integrating model-based decision support in a multi-modal reasoning system for managing type 1 diabetic patients

We present a multi-modal reasoning (MMR) methodology that integrates case-based reasoning (CBR), rule-based reasoning (RBR) and model-based reasoning (MBR), meant to provide physicians with a reliable decision support tool in the context of type 1 diabetes mellitus management. In particular, we have implemented a decision support system that is able to jointly exploit a probabilistic model of the glucose-insulin system at the steady state, a RBR system for suggestion generation and a CBR system for patient's profiling. The integration of the CBR, RBR and MBR paradigms allows for an optimized exploitation of all the available information, and for the definition of a therapy properly tailored to the patient's needs, overcoming the single approaches limitations. The system has been tested both on simulated and on real patients' data.

A review of evidence of health benefit from artificial neural networks in medical intervention

The purpose of this review is to assess the evidence of healthcare benefits involving the application of artificial neural networks to the clinical functions of diagnosis, prognosis and survival analysis, in the medical domains of oncology, critical care and cardiovascular medicine. The primary source of publications is PUBMED listings under Randomised Controlled Trials and Clinical Trials. The rĵle of neural networks is introduced within the context of advances in medical decision support arising from parallel developments in statistics and artificial intelligence. This is followed by a survey of published Randomised Controlled Trials and Clinical Trials, leading to recommendations for good practice in the design and evaluation of neural networks for use in medical intervention.

Telemedical devices in diabetes management

This paper reviews telemedicine and its recent expansions within diabetes management. Diabetes mellitus continues to be one of the major chronic diseases with up to 11% of national health care expenditure, when all late complications are taken into account. Over and above this, the incidence of diabetes is increasing in pandemic fashion. Diabetes has been the focus of telemedicine and information technology over the past two decades. Useful applications supporting high quality treatment exist in clinical management, education, decision support and modelling. Development of high-speed networks enables transmission of good quality photographs making consultations from distant locations possible. Databases and data analysis are fundamental to diabetes outcome research. Less successful has been the development of electronic medical records although this is a dream of many. Evidence of improved clinical outcome using telemedical applications still awaits the breakthrough.

Impaired absorption and omission of insulin: a novel method of detection using the diabetes advisory system computer model

The Diabetes Advisory System (DIAS) is a decision-support program developed to assist insulin dose adjustment in type 1 diabetes. In this paper, we show how it might be used to identify impaired absorption or omission of insulin in patients with poorly controlled blood glucose. An evaluation of glucose results from four outpatients with persistent hyperglycemia is presented (age 19-48 years with type 1 diabetes for 13-18 years of duration, HbA1c 9.4-13.6%). Each had completed a 4-day record of blood glucose (BG, pre-meal and bedtime), dietary (carbohydrate) intake, and insulin doses (with injection sites). From these data, DIAS modeled a glucose profile (simulated glucose, SG) for the same period. Qualitative assessments were made of differences between BG and SG, and selective reduction or complete removal of insulin doses where BG >> SG. Large improvements in modeling were attributed to either impaired absorption or omission of insulin. Confirmation of these problems was sought from the patients by detailed consultation and physical examination. Impaired insulin absorption was suspected in two patients, both having significant injection site abnormalities. Insulin omission was suspected in the other two subjects. Both had normal injection sites, and one admitted to missing doses. Following retraining, data from three patients showed noticeable improvements in overall modeling as well as glucose control. Using DIAS in the evaluation of patients with type 1 diabetes may highlight previously unrecognized injection site abnormalities or insulin dose omission. This could assist rational optimization of insulin therapy in cases of persistently poor glucose control.

Intelligent analysis of clinical time series: an application in the diabetes mellitus domain

This paper describes the application of a method for the intelligent analysis of clinical time series in the diabetes mellitus domain. Such a method is based on temporal abstractions and relies on the following steps: (i) 'pre-processing' of raw data through the application of suitable filtering techniques: (ii) 'extraction' from the pre-processed data of a set of abstract episodes (temporal abstractions); and (iii) 'post-processing' of temporal abstractions; the post-processing phase results in a new set of features that embeds high level information on the patient dynamics. The derived features set is used to obtain new knowledge through the application of machine learning algorithms. The paper describes in detail the application of this methodology and presents some results obtained on simulated data and on a data-set of four diabetic patients monitored for > 1 year.

A model of the endogenous glucose balance incorporating the characteristics of glucose transporters

This paper describes the development and preliminary test of a model of the endogenous glucose balance that incorporates the characteristics of the glucose transporters GLUT1, GLUT3 and GLUT4. In the modeling process the model is parameterized with nine parameters that are subsequently estimated from data in the literature on the hepatic- and endogenous- balances at various combinations of blood glucose and insulin levels. The ability of the resulting endogenous balance to fit blood glucose measured from patients was tested on 20 patients. The fit obtained with this model compared favorably with the fit obtained with the endogenous balance currently incorporated in the DIAS system.

Models of subcutaneous insulin kinetics. A critical review

Subcutaneous insulin kinetics is a complex process whose quantitation is needed for a reliable glycemic control in the conventional therapy of insulin-dependent diabetes. The major difficulties in modeling include accounting for the distribution in the subcutaneous depot and transport to plasma. A single model describing in detail the various processes for all the commercially available insulin preparations is not available. Several models however have been proposed which vary in the degree of complexity. Virtually all of them handle the regular insulin preparation while a few handle the intermediate acting and the novel insulin analogues. In this paper we critically review these models.

Bayesian analysis of blood glucose time series from diabetes home monitoring

This paper describes the application of a novel Bayesian estimation technique to extract the structural components, i.e., trend and daily patterns, from blood glucose level time series coming from home monitoring of insulin dependent diabetes mellitus patients. The problem is formulated through a set of stochastic equations, and is solved in a Bayesian framework by using a Markov chain Monte Carlo technique. The potential of the method is illustrated by analyzing data coming from the home monitoring of a 14-year old male patient.

The promise of technology in the management of the patient with diabetes

The advent of technology has brought many improvements in the management of individual aspects of the care of the patient with diabetes. However, the best management requires communication between systems to enable the clinician to coordinate these various aspects. This article reviews examples of the application of technology to the individual aspects of care. It also discusses the problems and promise of technology to improve overall care management.

A neural network approach for insulin regime and dose adjustment in type 1 diabetes

BACKGROUND

A decision support system based on a neural network approach is proposed to advise on insulin regime and dose adjustment for type 1 diabetes patients.

METHOD

The system consists of two feed-forward neural networks, trained with the back-propagation algorithm with momentum and adaptive learning rate. The input to the system consists of patient's glucose levels, insulin intake, and observed hypoglycemia symptoms during a short time period. The output of the first neural network provides the insulin regime, which is applied as input to the second neural network to estimate the appropriate insulin doses for a short time period.

RESULTS

The system's ability in order to recommend on insulin regime is excellent, while its performance in adjusting the insulin dosages for a specific patient is highly dependent on the data set used during the training procedure.

CONCLUSIONS

Despite the limitations of computer-based approaches, this study shows that artificial neural networks can assist diabetes patients in insulin adjustment.