IFAC-PapersOnline

Front cover

04:00 PM Wednesday 31, 1969

Welcome and Introduction

04:00 PM Wednesday 31, 1969

Technical Program

04:00 PM Wednesday 31, 1969

Author Index

04:00 PM Wednesday 31, 1969

Keyword Index

04:00 PM Wednesday 31, 1969

Tight Glycemic Control - the Leading Role of Insulin Sensitivity in Determining Efficacy and Thus Outcome

04:00 PM Wednesday 31, 1969

Tight glycaemic control (TGC) has emerged as a major focus in critical care. However, repeating the initial successful reductions in reducing mortality and other outcomes via TGC has proven very difficult. Hence, there has been growing debate over the necessity of TGC, its goals, safety from hypoglycemia, and target cohorts. This article reviews existing knowledge and results to provide a new interpretation and explanation for the variable results in applying TGC. It then uses a validated metabolic system model to show how the root cause is the intra- and inter- patient variability, which makes TGC difficult over diverse cohorts and thus yields such variable results over many protocols.

Prediction Performance Comparison between Three Intensive Care Unit Glucose Models

04:00 PM Wednesday 31, 1969

In this paper the prediction performance of two models that were particularly developed for predicting the blood glucose signal in critically ill patients, and a third (rather naive) model were compared. The imposed real-life conditions were challenging as the prediction processes started at time step 1 (comparable to the admission of a patient at the intensive care unit) and the prediction horizon was set at 4 hours (although accurate prediction of the blood glucose signal in the initial phase after admission is difficult due to lack of patient-specific data). The results of one of the models was satisfactory in terms of forecasting ability and showed its potential to be validated for use in a predictive control system in real-life.

Development of a Model-Based Clinical Sepsis Biomarker for Critically Ill Patients

04:00 PM Wednesday 31, 1969

Sepsis occurs frequently in the intensive care unit (ICU) and is a leading cause of admission, mortality, and cost. Treatment guidelines recommend early intervention, however positive blood culture results may take up to 48 hours. Insulin sensitivity (SI) is known to decrease with worsening condition and could thus be used to aid diagnosis. Some glycemic control protocols are able to accurately identify insulin sensitivity in real-time. Receiver operator characteristic (ROC) curves and cut-off SI values for sepsis diagnosis were calculated for real-time model-based insulin sensitivity from glycemic control data of 36 patients with sepsis. Patients were identified as having sepsis based on a clinically validated sepsis score (ss) of 2 or higher (ss = 0-4 for increasing severity). A clinical biomarker was calculated from patient clinical data to maximize the discrimination between cohorts. Insulin sensitivity as a sepsis biomarker for diagnosis of severe sepsis achieves a 50% sensitivity, 76% specificity, 4.8% PPV, and 98.3% NPV at a SI cut-off value of 0.00013 L*mU min-1. A clinical biomarker combining SI, temperature, heart rate, respiratory rate, blood pressure, and their respective hourly rates of change achieves 73% sensitivity, 80% specificity, 8.4% PPV, and 99.2% NPV. Thus, a clinical biomarker provides an effective real-time negative predictive diagnostic for severe sepsis. Examination of both inter- and intra-patient statistical distribution of this biomarker and sepsis score show potential avenues to improve the positive predictive value.

Blood Glucose Control in Neonatal Intensive Care with Model-Based Controllers

04:00 PM Wednesday 31, 1969

Premature neonates often experience hyperglycaemia, which has been linked to increased mortality and worsened outcomes. Insulin therapy can assist in controlling blood glucose levels. However a reliable, robust control protocol is required to avoid hypoglycaemia and to meet nutrition goals. This study presents an adaptive, model-based predictive controller designed to incorporate the unique metabolic state and control parameters of the neonate. Controller performance was tested in virtual trials on a 25 patient retrospective cohort and 24-hour pilot clinical trials. The effects of measurement frequency and BG sensor error were also evaluated. Time in the 4 7 mmol/L BG band was increased by 110%-145% compared to retrospective control for that cohort, with fewer hypoglycaemic measurements. Controllers were robust to BG sensor errors.

Corticosteroids and Insulin Resistance in the ICU

04:00 PM Wednesday 31, 1969

Corticosteroids reduce insulin sensitivity in healthy individuals by 30- 62-percent. The aim of this research was to use model-based methods to determine whether this reduction is also true in critically ill patients and how it may affect tight glycaemic control. A clinically validated model-based measure of insulin sensitivity was used to quantify changes between two matched cohorts of 40 intensive care unit (ICU) patients from Christchurch hospital. A 9-percent reduction in median insulin sensitivity was seen between the control cohort and patients receiving corticosteroids (per patient dose equivalent to 160mg/d of hydrocortisone). On a per-patient basis 11- 22-percent reductions were observed with higher percentile patients having greater suppression of insulin sensitivity. This research has shown that corticosteroids cause a much lower reduction in insulin sensitivity for critically ill patients compared to healthy controls and may thus have far less impact than suspected on glycaemic control in the ICU setting.

Physiologic Insulin Delivery with Insulin Feedback: A Control Systems Perspective

04:00 PM Wednesday 31, 1969

Individuals with type 1 diabetes mellitus must keep good control of their glucose levels in order to avoid the complications associated with the disease. This is no easy task, however, since even a highly motivated patient can have a hard time doing so. Moreover, this imposes a significant burden, as patients must be constantly monitoring and adjusting their treatment. For this reason, research into a closed-loop insulin delivery system has been of interest for several decades. This paper provides an overview, from a control systems perspective, of the research and development effort of a particular algorithm the external Physiologic Insulin Delivery system. In particular the introduction of insulin feedback, as suggested by beta-cell physiology, is covered in detail. A summary of human clinical trials is provided in the context of the evolution of this algorithm, and this paper outlines some of the research avenues that show particular promise.

Anticipating Meals with Behavioral Profiles: Towards Stochastic Model Predictive Control of T1DM

04:00 PM Wednesday 31, 1969

The delay associated with subcutaneous glucose sensing and insulin infusion actuation significantly complicates the design of control algorithms for regulating blood glucose in patients with Type 1 Diabetes Mellitus (T1DM). Model predictive control (MPC) is one strategy for mitigating delay, where optimal insulin infusions can be given in anticipation of future meal disturbances. Unfortunately, exact prior knowledge of meals can only be assured in a clinical environment, and uncertainty about when and if meals will arrive could lead to catastrophic outcomes. In this paper we develop an MPC-like control law that can anticipate meals given a probabilistic description of the patient's eating behavior in the form of a random meal (behavioral) profile. Preclinical in silico trials using the oral glucose meal model of Dalla Man et al. show that the control strategy provides a convenient means to account for uncertain prior knowledge of meals without compromising patient safety, even in the event that anticipated meals are skipped.

Model-Based Decision Support System to Improve Diabetes Care and Management

04:00 PM Wednesday 31, 1969

The model-based Karlsburg Diabetes Management System (KADIS) has been developed as a decision support system for physicians in their efforts to optimize metabolic control in diabetes care. For this purpose, KADIS was evaluated under different conditions by conducting open-label mono and polycentric trials, a case-control study and, last but not least, an observational study in routine diabetes care. The trial outcomes clearly show that the recommendations provided to the physicians by KADIS lead to significant improvement of metabolic control. It is concluded that the model-based decision support system provides an excellent tool to effectively guide physicians in decision making to achiev optimal metabolic control for their patients.

The Effect of Glargine As Basal Insulin Support for Recovering Critically Ill and High Dependency Unit Patients: An in Silico Study

04:00 PM Wednesday 31, 1969

Many critically ill patients are benefiting from extensive research done in tight glucose control (TGC) within the ICU. But moderate to high levels of hyperglycaemia are still tolerated within high dependency (HDU) and surgical units. The use and benefits of insulin protocols within these units have not yet been addressed in the literature. The management of tight glycaemic control still remains under the influence of ineffective standards characterized by tolerance for hyperglycaemia and a reluctance to use insulin intensively. A validated Glargine and intravenous insulin-glucose pharmacodynamic model are presented. Virtual trial results on 16 stable ICU patients showed that Glargine can provide effective blood glucose management for these long term recovering patients. An initial intravenous injection and higher Glargine dosing is required for the first day to quickly lower elevated blood glucose levels. However, once patients blood glucose levels are within a desirable range, Glargine alone can provide effective glycaemic management, thus reducing nursing effort. Median blood glucose for the entire cohort when simulated with the combination of Glargine and an intravenous insulin injection is 6.5 with interquartile range of [5.6, 7.5]. The 90% confidence interval is [4.6, 9.7] with no occurrence of hypoglycaemia. This in silico study provides a first virtual trial analysis of the in-hospital transition between intravenous and subcutaneous insulin for TGC.

Induced L2-Norm Minimization of Glucose-Insulin System for Type I Diabetic Patients

04:00 PM Wednesday 31, 1969

Using induced L2-norm minimization, a robust controller was developed for insulin delivery in Type I diabetic patients. The high-complexity nonlinear diabetic patient Sorensen-model was considered and Linear Parameter Varying methodology was used to develop open loop model and robust controller. Considering the normoglycemic set point (81.1 mg/dL), a polytopic set was created over the physiologic boundaries of the glucose-insulin interaction of the Sorensen-model. In this way, Linear Parameter Varying model formalism was defined. The robust control was developed considering input and output multiplicative uncertainties with two additional uncertainties from those used in the literature: sensor noise and worst case design for meal disturbance (60 g carbohydrate).

Contrast-Enhanced Ultrasound Imaging of Insulin-Induced Microvascular Recruitment in Type 1 Diabetes

04:00 PM Wednesday 31, 1969

Type 1 diabetes mellitus (T1DM) patients relies exclusively on exogenous insulin to maintain safe blood glucose levels. It has been recently demonstrated that insulin promotes its own action by enhancing capillary recruitment which occurs 15-20 minutes before increase in glucose uptake. In this study, we examine insulin-induced capillary recruitment in T1DM patients with various levels of insulin sensitivity. Eleven hyperinsulinemic euglycemic clamps were performed on nine T1DM patients during which capillary recruitment was assessed at basal insulin concentration levels and at hyperphysiological levels 30 min after the start of the clamp using contrast-enhance ultrasound (CEU) imaging. A systematic procedure was developed to select regions of interest (ROI) in the sequences of ultrasound images, thus bypassing the subjectivity of a selection done ''by hand''. The replenishment curves were then obtained and fitted with exponential curves which parameters provide measures of microvascular blood volume (MBV) and microvascular flow (MF). Higher basal MBV and MF were observed in high insulin sensitive subjects compared to low SI subjects. Hyperphysiological insulin concentrations induced by the clamp yielded opposite effects in low insulin sensitivity (SI) and high SI patients: capillary recruitment and derecruitment was observed respectively. It was hypothesized that insulin-induced capillary recruitment occurs at a subject-specific level of insulin concentration, and that this level is low (below basal) for sensitive patients and higher (above basal) for more resitant patients. Furthermore, diabetic microvascular complications significantly correlated with the effect of hyperphysiological insulin levels on the capillary recruitment/derecruitment in the two groups.

Comparison of Identification Methods of a Time-Varying Insulin Sensitivity Parameter in a Simulation Model of Glucose Metabolism in the Critically Ill

04:00 PM Wednesday 31, 1969

Models of glucose metabolism can help to simulate and predict the blood glucose response in hyperglycaemic, critically ill patients. Model prediction performance depends on a sufficiently accurate estimation of the patient's time-varying insulin sensitivity. The work presents three least squares approaches, the integral method and a Bayesian method that have been compared by prediction accuracy on an absolute and on a relative scale. Clinical data yields 1491 blood glucose predictions based on 10 critically ill patients. The Bayesian approach proved to be best with small errors (9.7% absolute percent error, 14.7 root mean square of logarithmic error for prediction times <= 2h), and fewer and smaller outliers compared to the other methods. Computationally, the Bayesian method took 1.5 times longer per prediction compared to the fastest method. It can be concluded that a Bayesian parameter estimation gives safe and effective results for the insulin sensitivity estimation for this model.

Glucose Control in Critically Ill Patients Using Sliding Mode Control with Robust Differentiators

04:00 PM Wednesday 31, 1969

This paper presents a sliding mode controller for blood glucose control in critically ill patients. Most of the critically ill patients require an insulin infusion to regulate the elevated glucose levels mainly produced by counter-regulatory hormone secretion and insulin resistance. Sliding mode control can successfully handle problems with parameter uncertainties and nonlinearities. Robust differentiators have been incorporated in order to approximate the derivatives that appear in the resulting control law, and to avoid the abrupt changes in the control signal that are produced by the presence of noise and set point changes. The controller has been tested in silico with a literature model. Validation scenarios include parameter uncertainties, sensor noise and set point changes.

A New General Glucose Homeostatic Model Using a Proportional-Integral-Derivative Controller

04:00 PM Wednesday 31, 1969

The glucose-insulin system is a challenging process to model due to the feedback mechanisms present, hence the implementation of a model-based approach to the system is an on-going and challenging research area. A new approach is proposed here which provides an effective way of characterising glycaemic regulation. The resulting model is built on the premise that there are three phases of insulin secretion, similar to those seen in a proportional- integral-derivative (PID) type controller used in engineering control problems. The model relates these three phases to a biological understanding of the system, as well as the logical premise that the homeostatic mechanisms will maintain very tight control of the system. It includes compartments for insulin, glucose, insulin action and a compartment to simulate an integral function of glucose. Structural identifiability analysis was performed on the model to determine whether a unique set of parameter values could be obtained from the available observations, which would allow meaningful conclusions to be drawn from parameter estimation. Although two parameters - glucose production rate and the proportional control coefficient - were found to be unidentifiable, the former is not a concern as this is known to be impossible to measure without a tracer experiment, and the latter can be easily estimated from other means. Subsequent parameter estimation and the model simulations have shown good agreement with respect to real data.

A Sliding Mode Predictive Control Approach to Closed-Loop Glucose Control for Type 1 Diabetes

04:00 PM Wednesday 31, 1969

The development of robust and efficient glucose control algorithms is key to making the artificial pancreas a reality. In this paper a sliding mode predictive control (SMPC) is obtained by combining the design technique of a sliding mode control with a model based predictive control (MPC). The SMPC combines the main advantages of the two control methods: the robust features of SMC and the good performance of MPC, including the handling of constraints on manipulated and controlled variables. Control action is composed by three parts: a predictive one from an optimization problem, a discontinuous one given by the switching term and finally, a feed-forward action given by an insulin bolus that is injected when a meal is ingested. The prediction model is linear and it is represented by a second order model with time delay. In order to test the controller in silico experiments, the Hovorka model has been considered. The proposed control algorithm shows considerable robustness for intra-patient variability, as well as an enhanced ability to handle measurement uncertainties and disturbance rejection, especially focusing on postprandial behaviour.

DISTq: Low-Cost, Accurate and Real-Time Estimation of Insulin Sensitivity

04:00 PM Wednesday 31, 1969

Knowing insulin sensitivity (SI) can optimise glycaemic control, assess metabolic drug therapy, or define diabetes risk. The DISTq is a short, low dose IM-IVGTT that generates an estimate of SI immediately after a 40 minute test using only glucose measurements, subjects physical attributes, and population parameter estimations. In this article, the DISTq is evaluated in clincial and in silics trials. In clinical trials, the test has shown a very strong correlation to the fully sampled DIST SI (R=0.91), (which also uses insulin and c-peptide assays) and a strong correlation to the euglycemic hyperinsulinaemic clamp (EIC) in in silico virtual trials (R=0.81). This study shows that population estimates can reduce the need for expensive insulin and c-peptide assays in obtaining an accurate, realtime estimation of SI.

Structural Identifiability of the Minimal Model and a Euglycemic Hyperinsulinemic Clamp Model for Glucose-Insulin Dynamics

04:00 PM Wednesday 31, 1969

Many mathematical models have been developed to describe glucose and insulin kinetics as a means of analysing the effective control of diabetes. Of these probably the most widely accepted is the so-called Minimal Model. This paper concentrates on structural identifiability analyses of two well-established mathematical models that have been developed to characterise glucose-insulin kinetics under different experimental scenarios. Such analysis is a pre-requisite to experiment design and parameter estimation and is applied for the first time to these models with the structures considered. The analysis is initially applied to a basic form of the Minimal Model using the Taylor Series approach, demonstrating global identifiability contrary to previously published results. A now well-accepted extended form of the Minimal model is then considered where the model is provided in an augmented form in order to apply the Similarity Transformation approach. This augmentation was necessary to cater for the particular structure for the Minimal Model that includes a time term. Once more the system, proved to be globally identifiable when both glucose and insulin are observed. Due to the inappropriate nature of the Minimal Model with regard to glucose clamping an alternative model describing the glucose-insulin dynamics during Euglycemic Hyperinsulinemic Clamp was considered. The structural identifiability analysis of the Euglycemic Hyperinsulinemic Clamp model is also performed using the Taylor Series Approach and results show that, with glucose infusion as input alone, the model is structurally globally identifiable.

Indistinguishability and Identifiability of Kinetic Models for the Mur C Reaction in Peptidoglycan Biosynthesis

04:00 PM Wednesday 31, 1969

An important question in Systems Biology is the design of experiments to allow discrimination between two (or more) competing pathway models or biological mechanisms. In chemical kinetics a common assumption when studying reactions which release several products is to assume that they are all released in one step. A structural indistinguishability analysis is performed between two different models describing the kinetic mechanism of the Mur C reaction in the cytoplasmic phase of peptidoglycan biosynthesis. One model involves ordered substrate binding and ordered release of the three products; the competing model also assumes ordered substrate binding, but with fast release of the three products. The two versions are shown to be distinguishable both in the full version and under quasi-steady-state assumptions. A structural identifiability analysis is carried out for both models to ensure that the model output uniquely determines the unknown parameters. Similar analyses (indistinguishability and identifiability) are performed using other model simplifications (using conservation equations) and comparisons made with the results of the full model. The analysis forms an integrated step towards the modelling of the full pathway of the cytoplasmic phase of peptidoglycan biosynthesis.

Kinetic Modelling of the Role of the Aldehyde Dehydrogenase Enzyme and the Breast Cancer Resistance Protein in Drug Resistance and Transport

04:00 PM Wednesday 31, 1969

A compartmental model for the in vitro uptake kinetics of the anti-cancer agent topotecan (TPT) has been extended from a previously published model. The extended model describes the drug activity and delivery of the pharmacologically active form to the DNA target as well as the catalysis of the aldehyde dehydrogenase (ALDH) enzyme and the elimination of drug from the cytoplasm via the active pump. Verification of the proposed model is achieved using scanning-laser microscopy data from live human breast cancer cells. Before estimating the unknown model parameters from the collected data it is essential to determine parameter uniqueness (or otherwise) from this imposed output structure. This is formally performed as a structural identifiability analysis, which demonstrates that all of the unknown model parameters are uniquely determined by the output structure corresponding to the real experiment.

Compartmental Modelling of the Pharmacokinetics of a Breast Cancer Resistance Protein

04:00 PM Wednesday 31, 1969

A mathematical model for the pharmacokinetics of Hoechst 33342 following administration into a culture medium containing a population of transfected cells (HEK293 hBCRP) with a potent inhibitor, Fumitremorgin C (FTC), present is described. This non-linear compartmental model has seven macroscopic sub-units, with fourteen rate parameters. A kinetic modelling software package, namely FACSIMILE (MPCA Software, UK), was used to obtain numerical solutions for the system equations and for parameter fitting. Model fits gave good agreement with in-vitro data provided by AstraZeneca.