The Lung Physiome and virtual patient models: From morphometry to clinical translation.

The understanding or prediction of specific functions of the lung can be made using compact models that have identifiable parameters and that are custom designed to the problem of interest. However, when structure contributes to function - as is the case with most lung pathologies - structure-based, biophysical models become essential. Here we describe the application of structure-based models within the lung Physiome framework to identifying and explaining patient risk in 12patients diagnosed with acute pulmonary embolism. The model integrates perfusion, ventilation, and gas exchange to predict arterial blood gases and pulmonary artery pressure in individual patient models in response to patient-specific blood clot distribution, with full or partial arterial occlusion. The necessity for a patient-specific approach with biophysical models that account for scale-specific structure and function is demonstrated.

Cheek support affects lung mechanics measurements of tidal-based spontaneous breathing.

BACKGROUND AND OBJECTIVE: Patients are required to support their cheeks during breath-occluding lung function tests. This prevents cheek expansion which would alter pressure measured at the mouth, and, consequently, lung mechanics measurements. To date, the effect of cheek support on airway resistance measurements has been assessed. However other lung mechanics have not been studied as thoroughly, and no algorithm to account for the effect of missing cheek support on lung mechanics measurements has been developed. METHODS: Lung mechanics were assessed with a breath occlusion test during light panting in healthy subjects with and without cheek support in a body plethysmograph. Average model-based airway resistance, lung elastance, and a parameter representing the viscoelastic were measured. Results were compared to quantify the effect of cheek support on these three parameters. RESULTS: In the nine healthy subjects (5 Female, 4 Male) recruited for this study, all mechanics tended to be underestimated when cheeks were unsupported. Changes in elastance, resistance, and viscoelastic parameter ranged between 1.6-66.8 %, -4.5-21.8 %, and -4.7-68.2 %, respectively, when cheek support was added. The underestimation was due to reduced mouth pressure during cheek expansion when the breath was occluded. The variance of lung mechanics parameters did not change with cheek support in all subjects. CONCLUSIONS: The error in lung mechanics measurement caused by unsupported cheeks was subject dependent. Hence, no rule-of-thumb could be identified to reconstruct missing cheek support. For correct lung mechanics measurements during breath-occluding lung tests, patients must have adequate cheek support. ABBREVIATIONS: ROCC: Occlusion resistance; COPD: Chronic Obstructive Pulmonary Disorder; SB: spontaneous breathing.

Inspiratory respiratory mechanics estimation by using expiratory data for reverse-triggered breathing cycles.

BACKGROUND AND OBJECTIVE: Model-based lung mechanics monitoring can provide clinically useful information for guiding mechanical ventilator treatment in intensive care. However, many methods of measuring lung mechanics are not appropriate for both fully and partially sedated patients, and are unable provide lung mechanics metrics in real-time. This study proposes a novel method of using lung mechanics identified during passive expiration to estimate inspiratory lung mechanics for spontaneously breathing patients. METHODS: Relationships between inspiratory and expiratory modeled lung mechanics were identified from clinical data from 4 fully sedated patients. The validity of these relationships were assessed using data from a further 4 spontaneously breathing patients. RESULTS: For the fully sedated patients, a linear relationship was identified between inspiratory and expiratory elastance, with slope 1.04 and intercept 1.66. The r value of this correlation was 0.94. No cohort-wide relationship was determined for airway resistance. Expiratory elastance measurements in spontaneously breathing patients were able to produce reasonable estimates of inspiratory elastance after adjusting for the identified difference between them. CONCLUSIONS: This study shows that when conventional methods fail, typically ignored expiratory data may be able to provide clinicians with the information needed about patient condition to guide MV therapy.

Modelling intestinal glucose absorption in premature infants using continuous glucose monitoring data.

BACKGROUND: Model-based glycaemic control protocols have shown promise in neonatal intensive care units (NICUs) for reducing both hyperglycaemia and insulin-therapy driven hypoglycaemia. However, current models for the appearance of glucose from enteral feeding are based on values from adult intensive care cohorts. This study aims to determine enteral glucose appearance model parameters more reflective of premature infant physiology. METHODS: Peaks in CGM data associated with enteral milk feeds in preterm and term infants are used to fit a two compartment gut model. The first compartment describes glucose in the stomach, and the half life of gastric emptying is estimated as 20 min from literature. The second compartment describes glucose in the small intestine, and absorption of glucose into the blood is fit to CGM data. Two infant cohorts from two NICUs are used, and results are compared to appearances derived from data in highly controlled studies in literature. RESULTS: The average half life across all infants for glucose absorption from the gut to the blood was 50 min. This result was slightly slower than, but of similar magnitude to, results derived from literature. No trends were found with gestational or postnatal age. Breast milk fed infants were found to have a higher absorption constant than formula fed infants, a result which may reflect known differences in gastric emptying for different feed types. CONCLUSIONS: This paper presents a methodology for estimation of glucose appearance due to enteral feeding, and model parameters suitable for a NICU model-based glycaemic control context.

Model-based identification and diagnosis of a porcine model of induced endotoxic shock with hemofiltration.

A previously validated cardiovascular system (CVS) model and parameter identification method for cardiac and circulatory disease states are extended and further validated in a porcine model (N=6) of induced endotoxic shock with hemofiltration. Errors for the identified model are within 10% when the model is re-simulated and compared to the clinical data. All identified parameter trends over time in the experiments match clinically expected changes both individually and over the cohort. This work represents a further clinical validation of these model-based cardiovascular diagnosis and therapy guidance methods for use with monitoring endotoxic disease states.

Classifying algorithms for SIFT-MS technology and medical diagnosis.

Selected Ion Flow Tube-Mass Spectrometry (SIFT-MS) is an analytical technique for real-time quantification of trace gases in air or breath samples. SIFT-MS system thus offers unique potential for early, rapid detection of disease states. Identification of volatile organic compound (VOC) masses that contribute strongly towards a successful classification clearly highlights potential new biomarkers. A method utilising kernel density estimates is thus presented for classifying unknown samples. It is validated in a simple known case and a clinical setting before-after dialysis. The simple case with nitrogen in Tedlar bags returned a 100% success rate, as expected. The clinical proof-of-concept with seven tests on one patient had an ROC curve area of 0.89. These results validate the method presented and illustrate the emerging clinical potential of this technology.

Prediction of hemodynamic changes towards PEEP titrations at different volemic levels using a minimal cardiovascular model.

A cardiovascular system model and parameter identification method have previously been validated for porcine experiments of induced pulmonary embolism and positive end-expiratory pressure (PEEP) titrations, accurately tracking all the main hemodynamic trends. In this research, the model and parameter identification process are further validated by predicting the effect of intervention. An overall population-specific rule linking specific model parameters to increases in PEEP is formulated to predict the hemodynamic effects on arterial pressure, pulmonary artery pressure and stroke volume. Hemodynamic changes are predicted for an increase from 0 to 10 cm H(2)O with median absolute percentage errors less than 7% (systolic pressures) and 13% (stroke volume). For an increase from 10 to 20 cm H(2)O median absolute percentage errors are less than 11% (systolic pressures) and 17% (stroke volume). These results validate the general applicability of such a rule, which is not pig-specific, but holds over for all analyzed pigs. This rule enables physiological simulation and prediction of patient response. Overall, the prediction accuracy achieved represents a further clinical validation of these models, methods and overall approach to cardiovascular diagnosis and therapy guidance.

Model-based identification of PEEP titrations during different volemic levels.

A cardiovascular system (CVS) model has previously been validated in simulated cardiac and circulatory disease states. It has also been shown to accurately capture all main hemodynamic trends in a porcine model of pulmonary embolism. In this research, a slightly extended CVS model and parameter identification process are presented and validated in a porcine experiment of positive end-expiratory pressure (PEEP) titrations at different volemic levels. The model is extended to more physiologically represent the separation of venous and arterial circulation. Errors for the identified model are within 5% when re-simulated and compared to clinical data. All identified parameter trends match clinically expected changes. This work represents another clinical validation of the underlying fundamental CVS model, and the methods and approach of using them for cardiovascular diagnosis in critical care.

A virtual patient model for mechanical ventilation.

BACKGROUND AND OBJECTIVES: Mechanical ventilation (MV) is a primary therapy for patients with acute respiratory failure. However, poorly selected ventilator settings can cause further lung damage due to heterogeneity of healthy and damaged alveoli. Varying positive-end-expiratory-pressure (PEEP) to a point of minimum elastance is a lung protective ventilator strategy. However, even low levels of PEEP can lead to ventilator induced lung injury for individuals with highly inflamed pulmonary tissue. Hence, models that could accurately predict peak inspiratory pressures after changes to PEEP could improve clinician confidence in attempting potentially beneficial treatment strategies. METHODS: This study develops and validates a physiologically relevant respiratory model that captures elastance and resistance via basis functions within a well-validated single compartment lung model. The model can be personalised using information available at a low PEEP to predict lung mechanics at a higher PEEP. Proof of concept validation is undertaken with data from four patients and eight recruitment manoeuvre arms. RESULTS: Results show low error when predicting upwards over the clinically relevant pressure range, with the model able to predict peak inspiratory pressure with less than 10% error over 90% of the range of PEEP changes up to 12 cmH2O. CONCLUSIONS: The results provide an in-silico model-based means of predicting clinically relevant responses to changes in MV therapy, which is the foundation of a first virtual patient for MV.

Model-based cardiac diagnosis of pulmonary embolism.

A minimal cardiac model has been shown to accurately capture a wide range of cardiovascular system dynamics commonly seen in the intensive care unit (ICU). However, standard parameter identification methods for this model are highly non-linear and non-convex, hindering real-time clinical application. An integral-based identification method that transforms the problem into a linear, convex problem, has been previously developed, but was only applied on continuous simulated data with random noise. This paper extends the method to handle discrete sets of clinical data, unmodelled dynamics, a significantly reduced data set theta requires only the minimum and maximum values of the pressure in the aorta, pulmonary artery and the volumes in the ventricles. The importance of integrals in the formulation for noise reduction is illustrated by demonstrating instability in the identification using simple derivative-based approaches. The cardiovascular system (CVS) model and parameter identification method are then clinically validated on porcine data for pulmonary embolism. Errors for the identified model are within 10% when re-simulated and compared to clinical data. All identified parameter trends match clinically expected changes. This work represents the first clinical validation of these models, methods and approach to cardiovascular diagnosis in critical care.

Insulin kinetics and the Neonatal Intensive Care Insulin-Nutrition-Glucose (NICING) model.

BACKGROUND: Models of human glucose-insulin physiology have been developed for a range of uses, with similarly different levels of complexity and accuracy. STAR (Stochastic Targeted) is a model-based approach to glycaemic control. Elevated blood glucose concentrations (hyperglycaemia) are a common complication of stress and prematurity in very premature infants, and have been associated with worsened outcomes and higher mortality. This research identifies and validates the model parameters for model-based glycaemic control in neonatal intensive care. METHODS: C-peptide, plasma insulin, and BG from a cohort of 41 extremely pre-term (median age 27.2 [26.2-28.7] weeks) and very low birth weight infants (median birth weight 839 [735-1000] g) are used alongside C-peptide kinetic models to identify model parameters associated with insulin kinetics in the NICING (Neonatal Intensive Care Insulin-Nutrition-Glucose) model. A literature analysis is used to determine models of kidney clearance and body fluid compartment volumes. The full, final NICING model is validated by fitting the model to a cohort of 160 glucose, insulin, and nutrition data records from extremely premature infants from two different NICUs (neonatal intensive care units). RESULTS: Six model parameters related to insulin kinetics were identified. The resulting NICING model is more physiologically descriptive than prior model iterations, including clearance pathways of insulin via the liver and kidney, rather than a lumped parameter. In addition, insulin diffusion between plasma and interstitial spaces is evaluated, with differences in distribution volume taken into consideration for each of these spaces. The NICING model was shown to fit clinical data well, with a low model fit error similar to that of previous model iterations. CONCLUSIONS: Insulin kinetic parameters have been identified, and the NICING model is presented for glycaemic control neonatal intensive care. The resulting NICING model is more complex and physiologically relevant, with no loss in bedside-identifiability or ability to capture and predict metabolic dynamics.

Effective sample size estimation for a mechanical ventilation trial through Monte-Carlo simulation: Length of mechanical ventilation and Ventilator Free Days.

Randomised control trials have sought to seek to improve mechanical ventilation treatment. However, few trials to date have shown clinical significance. It is hypothesised that aside from effective treatment, the outcome metrics and sample sizes of the trial also affect the significance, and thus impact trial design. In this study, a Monte-Carlo simulation method was developed and used to investigate several outcome metrics of ventilation treatment, including 1) length of mechanical ventilation (LoMV); 2) Ventilator Free Days (VFD); and 3) LoMV-28, a combination of the other metrics. As these metrics have highly skewed distributions, it also investigated the impact of imposing clinically relevant exclusion criteria on study power to enable better design for significance. Data from invasively ventilated patients from a single intensive care unit were used in this analysis to demonstrate the method. Use of LoMV as an outcome metric required 160 patients/arm to reach 80% power with a clinically expected intervention difference of 25% LoMV if clinically relevant exclusion criteria were applied to the cohort, but 400 patients/arm if they were not. However, only 130 patients/arm would be required for the same statistical significance at the same intervention difference if VFD was used. A Monte-Carlo simulation approach using local cohort data combined with objective patient selection criteria can yield better design of ventilation studies to desired power and significance, with fewer patients per arm than traditional trial design methods, which in turn reduces patient risk. Outcome metrics, such as VFD, should be used when a difference in mortality is also expected between the two cohorts. Finally, the non-parametric approach taken is readily generalisable to a range of trial types where outcome data is similarly skewed.

A novel, model-based insulin and nutrition delivery controller for glycemic regulation in critically ill patients.

BACKGROUND: Critically ill patients are often hyperglycemic and insulin resistant, as well as highly dynamic. Tight glucose control has been shown to significantly reduce mortality in critical care. A physiological model of the glucose-insulin regulatory system is improved and used to develop an adaptive control protocol utilizing both nutritional and insulin inputs to control hyperglycemia. The approach is clinically verified in a critical care patient cohort. METHODS: A simple two-compartment model for glucose rate of appearance in plasma due to stepwise enteral glucose fluxes is developed and incorporated into a previously validated system model. A control protocol modulating intravenous insulin infusion and bolus, with an enteral feed rate, is developed, enabling tight and predictive glycemic regulation to preset targets. The control protocol is adaptive to patient time-variant effective insulin resistance. The model and protocol are verified in seven 10-h and one 24-h proof-of-concept clinical trials. Ethics approval was granted by the Canterbury Ethics Committee. RESULTS: Insulin requirements varied widely following acute changes in patient physiology. The algorithm developed successfully adapted to patient metabolic status and insulin sensitivity, achieving an average target acquisition error of 9.3% with 90.7% of all targets achieved within +/-20%. Prediction errors may not be distinguishable from sensor measurement errors. Large errors (>20%) are attributable to highly dynamic and unpredictable changes in patient condition. CONCLUSIONS: Tight, targeted stepwise regulation was exhibited in all trials. Overall, tight glycemic regulation is achieved in a broad critical care cohort with optimized insulin and nutrition delivery, effectively managing glycemia even with high effective insulin resistance.

3D models of blood flow in the cerebral vasculature.

The circle of Willis (CoW) is a ring-like arterial structure located in the base of the brain and is responsible for the distribution of oxygenated blood throughout the cerebral mass. To investigate the effects of the complex 3D geometry and anatomical variability of the CoW on the cerebral hemodynamics, a technique for generating physiologically accurate models of the CoW has been created using a combination of magnetic resonance data and computer-aided design software. A mathematical model of the body's cerebral autoregulation mechanism has been developed and numerous computational fluid dynamics simulations performed to model the hemodynamics in response to changes in afferent blood pressure. Three pathological conditions were explored, namely a complete CoW, a fetal P1 and a missing A1. The methodology of the cerebral hemodynamic modelling is proposed with the potential for future clinical application in mind, as a diagnostic tool.

Physiological modelling of agitation-sedation dynamics including endogenous agitation reduction.

Sedation administration and agitation management are fundamental activities in any intensive care unit. A lack of objective measures of agitation and sedation, as well as poor understanding of the underlying dynamics, contribute to inefficient outcomes and expensive healthcare. Recent models of agitation-sedation pharmacodynamics have enhanced understanding of the underlying dynamics and enable development of advanced protocols for semi-automated sedation administration. However, these initial models do not capture all observed dynamics, particularly periods of low sedative infusion. A physiologically representative model that incorporates endogenous agitation reduction (EAR) dynamics is presented and validated using data from 37 critical care patients. High median relative average normalised density (RAND) values of 0.77 and 0.78 support and minimum RAND values of 0.51 and 0.55 for models without and with EAR dynamics respectively show that both models are valid representations of the fundamental agitation-sedation dynamics present in a broad spectrum of intensive care unit (ICU) patients. While the addition of the EAR dynamic increases the ability of the model to capture the observed dynamics of the agitation-sedation system, the improvement is relatively small and the sensitivity of the model to the EAR dynamic is low. Although this may represent a limitation of the model, the inclusion of EAR is shown to be important for accurately capturing periods of low, or no, sedative infusion, such as during weaning prior to extubation.

Physiological modelling 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. A model that captures the essential dynamics of the agitation-sedation system and is physiologically representative is developed, and validated using data from 37 critical care patients. It is more physiologically representative than a previously published agitation-sedation model, and captures more realistic and complex dynamics. The median time in the 90% probability band is 90%, and the total drug dose, relative to recorded drug dose data, is a near ideal 101%. These statistical model validation metrics are 5-13% better than a previously validated model. Hence, this research provides a platform to develop and test semi-automated sedation management controllers that offer the significant clinical potential of improved agitation management and reduced length of stay in critical care.

Model predictive glycaemic regulation in critical illness using insulin and nutrition input: a pilot study.

Stress-induced hyperglycaemia is prevalent in intensive care, impairing the immune response. Nutritional support regimes with high glucose content further exacerbate the problem. Tight glucose control has been shown to reduce mortality by up to 43% if levels are kept below 6.1 mmol/L. This research develops a control algorithm with insulin and nutritional inputs for targeted glucose control in the critically ill. Ethics approval for this research was granted by the Canterbury Ethics Committee. Proof-of-concept clinical pilot trials were conducted on intubated, insulin-dependent Christchurch ICU patients (n=7) on constant nutritional support. A target 10-15% reduction in glucose level per hour for a desired glucose level of 4-6 mmol/L was set. 43% and 91% of glucose targets were achieved within +/-5 and +/-20%, respectively. The mean error was 8.9% (0.5 mmol/L), with an absolute range [0, 2.9] mmol/L. End glucose levels were 40% lower compared to initial values. All large target errors are attributable to sudden changes in patient physiology at low glucose values, rather than systemic deficiencies. Target errors are consistent with and explainable by published sensor error distributions. The results show that intensive model-based glucose management with nutrition control reduced absolute glucose levels progressively while reducing the severity of glycaemic fluctuation even with significant inter-patient variability and time-varying physiological condition. Trials spanning longer periods of time are in development to verify the short-term pilot studies performed and to test the adaptability of the controller. Clinically, these results indicate potential in clinical use to reduce ICU mortality as well as reduce risk of severe complications.

Integral-based identification of patient specific parameters for a minimal cardiac model.

A minimal cardiac model has been developed which accurately captures the essential dynamics of the cardiovascular system (CVS). However, identifying patient specific parameters with the limited measurements often available, hinders the clinical application of the model for diagnosis and therapy selection. This paper presents an integral-based parameter identification method for fast, accurate identification of patient specific parameters using limited measured data. The integral method turns a previously non-linear and non-convex optimization problem into a linear and convex identification problem. The model includes ventricular interaction and physiological valve dynamics. A healthy human state and four disease states, valvular stenosis, pulmonary embolism, cardiogenic shock and septic shock are used to test the method. Parameters for the healthy and disease states are accurately identified using only discretized flows into and out of the two cardiac chambers, the minimum and maximum volumes of the left and right ventricles, and the pressure waveforms through the aorta and pulmonary artery. These input values can be readily obtained non-invasively using echo-cardiography and ultra-sound, or invasively via catheters that are often used in Intensive Care. The method enables rapid identification of model parameters to match a particular patient condition in clinical real time (3-5 min) to within a mean value of 4-10% in the presence of 5-15% uniformly distributed measurement noise. The specific changes made to simulate each disease state are correctly identified in each case to within 10% without false identification of any other patient specific parameters. Clinically, the resulting patient specific model can then be used to assist medical staff in understanding, diagnosis and treatment selection.

Fast normalized cross correlation for motion tracking using basis functions.

Digital image-based elasto-tomography (DIET) is an emerging method for non-invasive breast cancer screening. Effective clinical application of the DIET system requires highly accurate motion tracking of the surface of an actuated breast with minimal computation. Normalized cross correlation (NCC) is the most robust correlation measure for determining similarity between points in two or more images providing an accurate foundation for motion tracking. However, even using fast Fourier transform (FFT) methods, it is too computationally intense for rapidly managing several large images. A significantly faster method of calculating the NCC is presented that uses rectangular approximations in place of randomly placed landmark points or the natural marks on the breast. These approximations serve as an optimal set of basis functions that are automatically detected, dramatically reducing computational requirements. To prove the concept, the method is shown to be 37-150 times faster than the FFT-based NCC with the same accuracy for simulated data, a visco-elastic breast phantom experiment and human skin. Clinically, this approach enables thousands of randomly placed points to be rapidly and accurately tracked providing high resolution for the DIET system.

A comparison between four techniques to measure cardiac output.

Cardiac output is an important variable when monitoring hemodynamic status. In particular, changes in cardiac output represent the goal of several circulatory management therapies. Unfortunately, cardiac output is very difficult to estimate, either in experimental or clinical settings. The goal of this work is to compare four techniques to measure cardiac output: pressure-volume catheter, aortic flow probe, thermodilution, and the PiCCO monitor. These four techniques were simultaneously used during experiments of fluid and endotoxin administration on 7 pigs. Findings show that, first, each individual technique is precise, with a relative coefficient of repeatability lower than 7 %. Second, 1 cardiac output estimate provided by any technique relates poorly to the estimates from the other 3, even if there is only small bias between the techniques. Third, changes in cardiac output detected by one technique are only detected by the others in 62 to 100 % of cases. This study confirms the difficulty of obtaining a reliable clinical cardiac output measurement. Therefore, several measurements using different techniques should be performed, if possible, and all such should be treated with caution.