G.M. Shaw

Blood Glucose Prediction Using Stochastic Modeling in Neonatal Intensive Care

Hyperglycemia is a common metabolic problem in premature, low-birth-weight infants. Blood glucose homeostasis in this group is often disturbed by immaturity of endogenous regulatory systems and the stress of their condition in intensive care. A dynamic model capturing the fundamental dynamics of the glucose regulatory system provides a measure of insulin sensitivity (SI). Forecasting the most probable future SI can significantly enhance real-time glucose control by providing a clinically validated/proven level of confidence on the outcome of an intervention, and thus, increased safety against hypoglycemia. A 2-D kernel model of SI is fitted to 3567 h of identified, time-varying SI from retrospective clinical data of 25 neonatal patients with birth gestational age 23 to 28.9 weeks. Conditional probability estimates are used to determine SI probability intervals. A lag-2 stochastic model and adjustments of the variance estimator are used to explore the bias-variance tradeoff in the hour-to-hour variation of SI. The model captured 62.6% and 93.4% of in-sample SI predictions within the (25th-75th) and (5th-95th) probability forecast intervals. This overconservative result is also present on the cross-validation cohorts and in the lag-2 model. Adjustments to the variance estimator found a reduction to 10%-50% of the original value provided optimal coverage with 54.7% and 90.9% in the (25th-75th) and (5th-95th) intervals. A stochastic model of SI provided conservative forecasts, which can add a layer of safety to real-time control. Adjusting the variance estimator provides a more accurate, cohort-specific stochastic model of SI dynamics in the neonate.

The dynamic insulin sensitivity and secretion test--a novel measure of insulin sensitivity.

The objective was to validate the methodology for the dynamic insulin sensitivity and secretion test (DISST) and to demonstrate its potential in clinical and research settings. One hundred twenty-three men and women had routine clinical and biochemical measurements, an oral glucose tolerance test, and a DISST. For the DISST, participants were cannulated for blood sampling and bolus administration. Blood samples were drawn at t = 0, 10, 15, 25, and 35 minutes for measurement of glucose, insulin, and C-peptide. A 10-g bolus of intravenous glucose at t = 5 minutes and 1 U of intravenous insulin immediately after the t = 15 minute sample were given. Fifty participants also had a hyperinsulinemic-euglycemic clamp. Relationships between DISST insulin sensitivity (SI) and the clamp, and both DISST SI and secretion and other metabolic variables were measured. A Bland-Altman plot showed little bias in the comparison of DISST with the clamp, with DISST underestimating the glucose clamp by 0.1·10(-2)·mg·L·kg(-1)·min(-1)·pmol(-1) (90% confidence interval, -0.2 to 0). The correlation between SI as measured by DISST and the clamp was 0.82; the c unit for the receiver operating characteristic curve analysis for the 2 tests was 0.96. Metabolic variables showed significant correlations with DISST SI and the second phase of insulin release. The DISST also appears able to distinguish different insulin secretion patterns in individuals with identical SI values. The DISST is a simple, dynamic test that compares favorably with the clamp in assessing SI and allows simultaneous assessment of insulin secretion. The DISST has the potential to provide even more information about the pathophysiology of diabetes than more complicated tests.

Simulating transient ventricular interaction using a minimal cardiovascular system model.

A minimal closed-loop cardiovascular system (CVS) model has been developed that can simulate ventricular interaction due to both direct interaction through the septum and series interaction through the circulation system. The model is used to simulate canine experiments carried out to study the transient response of the left ventricle due to changes in right ventricle pressures and volumes. The model-simulated trends in left and right ventricle pressures and volumes, septum deflection and arterial flow rates are compared with the experimental results. In spite of the limited physiological data available describing the animals, the model is shown to capture all the transient trends in the experimental data. This is the first known example of a physiological model that can capture all these trends. The model is then used to illustrate the separate effects of direct and series interactions independently. This study proves the value of this modelling method to be used in conjunction with experimental data for delineating and understanding the factors that contribute to ventricular dynamics.

Adaptive Bolus-Based Set-Point Regulation of Hyperglycemia in Critical Care

Critically ill patients are often hyperglycemic and extremely diverse in their dynamics. Consequently, fixed protocols and sliding scales can result in error and poor control. A two-compartment glucose-insulin system model that accounts for time-varying insulin sensitivity and endogenous glucose removal, along with two different saturation kinetics is developed and verified in proof-of-concept clinical trials for adaptive control of hyperglycemia. The adaptive control algorithm monitors the physiological status of a critically ill patient, allowing real-time tight glycemic regulation. The bolus-based insulin administration approach is shown to result in safe, targeted stepwise glycemic reduction for three critically ill patients.

Patient specific identification of the cardiac driver function in a cardiovascular system model

The cardiac muscle activation or driver function, is a major determinant of cardiovascular dynamics, and is often approximated by the ratio of the left ventricle pressure to the left ventricle volume. In an intensive care unit, the left ventricle pressure is usually never measured, and the left ventricle volume is only measured occasionally by echocardiography, so is not available real-time. This paper develops a method for identifying the driver function based on correlates with geometrical features in the aortic pressure waveform. The method is included in an overall cardiovascular modelling approach, and is clinically validated on a porcine model of pulmonary embolism. For validation a comparison is done between the optimized parameters for a baseline model, which uses the direct measurements of the left ventricle pressure and volume, and the optimized parameters from the approximated driver function. The parameters do not significantly change between the two approaches thus showing that the patient specific approach to identifying the driver function is valid, and has potential clinically.

Dynamic models of ARDS lung mechanics for optimal patient ventilation

Mechanical ventilation is often used to treat patients with acute respiratory distress syndrome (ARDS). However, the optimal setting is still controversial, and physicians often rely on experience and intuition. The purpose of this research is to develop a model of the essential lung mechanics to help determining the optimal ventilator setting in clinical situations. The model is a compilation of physiologically based mechanics parameters, which are adjustable to represent patient specific conditions. Further investigation improvements are required, however it shows good initial for eventual clinical use.

Clinical Validation of a Model-based Glycaemic Control Design Approach and Comparison to Other Clinical Protocols

Hyperglycaemia is prevalent in critical care and tight control can reduce mortality from 9-43% depending on the level of control and the cohort. This research presents a table-based method that varies both insulin dose and nutritional input to achieve tight control. The system mimics a previously validated model-based system, but can be used for long term, large patient number clinical evaluation. This paper evaluates this method in simulation using retrospective data and then compares clinical measurements over 15,000 patient hours to validate the models and development approach. This validation thus also validates the in silico comparison to the landmark clinical tight glycaemic control protocols. Overall, an average clinical glucose level is 5.9plusmn1.0 mmol/L, matching simulation, however the overall clinical glucose distribution is slightly tighter than that obtained in simulation, indicating that the retrospective virtual trial design approach is slightly conservative. Finally, the model based approach is shown to have tighter control than existing, more ad-hoc clinical approaches based on the simulation results that qualitatively match reported clinical results, but also show significant variation around the average levels obtained in both the hypo-and hyperglycaemic ranges

Physiologically-based minimal model of agitation-sedation dynamics

Agitation-sedation cycling in critically ill patients, characterized by oscillations between states of agitation and over-sedation, damages patient health and increases length of stay and cost. The model presented captures the essential dynamics of the agitation-sedation system, is physiologically representative, and is validated by accurately simulating patient response for 37 critical care patients. The model provides a platform to develop and test controllers that offer the potential of improved agitation management.

Long term verification of glucose-insulin regulatory system model dynamics

Hyperglycaemia in critically ill patients increases the risk of further complications and mortality. A long-term verification of a model that captures the essential glucose- and insulin-kinetics is presented, using retrospective data gathered in an intensive care unit (ICU). The model uses only two patient specific parameters, for glucose clearance and insulin sensitivity. The optimization of these parameters is accomplished through a novel integration-based fitting approach, and a piecewise linearization of the parameters. This approach reduces the non-linear, non-convex optimization problem to a simple linear equation system. The method was tested on long-term blood glucose recordings from 17 ICU patients, resulting in an average error of 7%, which is in the range of the sensor error. One-hour predictions of blood glucose data proved acceptable with an error range between 711%. These results verify the models ability to capture long-term observed glucose-insulin dynamics in hyperglycaemic ICU patients.

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%.