Model-based Protocols Research Articles

Virtual patient framework for the testing of mechanical ventilation airway pressure and flow settings protocol

Background and ObjectiveModel-based and personalised decision support systems are emerging to guide mechanical ventilation (MV) treatment for respiratory failure patients. However, model-based treatments require resource-intensive clinical trials prior to implementation. This research presents a framework for generating virtual patients for testing model-based decision support, and direct use in MV treatment. MethodsThe virtual MV patient framework consists of 3 stages: 1) Virtual patient generation, 2) Patient-level validation, and 3) Virtual clinical trials. The virtual patients are generated from retrospective MV patient data using a clinically validated respiratory mechanics model whose respiratory parameters (respiratory elastance and resistance) capture patient-specific pulmonary conditions and responses to MV care over time. Patient-level validation compares the predicted responses from the virtual patient to their retrospective results for clinically implemented MV settings and changes to care. Patient-level validated virtual patients create a platform to conduct virtual trials, where the safety of closed-loop model-based protocols can be evaluated. ResultsThis research creates and presents a virtual patient platform of 100 virtual patients generated from retrospective data. Patient-level validation reported median errors of 3.26% for volume-control and 6.80% for pressure-control ventilation mode. A virtual trial on a model-based protocol demonstrates the potential efficacy of using virtual patients for prospective evaluation and testing of the protocol. ConclusionThe virtual patient framework shows the potential to safely and rapidly design, develop, and optimise new model-based MV decision support systems and protocols using clinically validated models and computer simulation, which could ultimately improve patient care and outcomes in MV.

Stochastic Modelling of Respiratory System Elastance for Mechanically Ventilated Respiratory Failure Patients.

While lung protective mechanical ventilation (MV) guidelines have been developed to avoid ventilator-induced lung injury (VILI), a one-size-fits-all approach cannot benefit every individual patient. Hence, there is significant need for the ability to provide patient-specific MV settings to ensure safety, and optimise patient care. Model-based approaches enable patient-specific care by identifying time-varying patient-specific parameters, such as respiratory elastance, Ers, to capture inter- and intra-patient variability. However, patient-specific parameters evolve with time, as a function of disease progression and patient condition, making predicting their future values crucial for recommending patient-specific MV settings. This study employs stochastic modelling to predict future Ers values using retrospective patient data to develop and validate a model indicating future intra-patient variability of Ers. Cross validation results show stochastic modelling can predict future elastance ranges with 92.59 and 68.56% of predicted values within the 5–95% and the 25–75% range, respectively. This range can be used to ensure patients receive adequate minute ventilation should elastance rise and minimise the risk of VILI should elastance fall. The results show the potential for model-based protocols using stochastic model prediction of future Ers values to provide safe and patient-specific MV. These results warrant further investigation to validate its clinical utility.

Insulin Resistance in ICU Patients: Women Have Stronger Metabolic Response

Glycaemic control (GC) has been associated with improved outcomes in critically ill patients. However, inter- and intra- patient metabolic variability significantly increase the risk of hypoglycaemia when using insulin to control glycaemia. Model-based protocols often identify key physiological parameters from patient data, and demonstrated safe and effective GC. Based on recent studies showing gender difference in insulin secretion, this study uses retrospective data to identify whether there exists a difference in sexes in metabolic stress response, and thus in how personalised GC is given.Retrospective data from 145 ICU patients under GC who started GC in the first 12 hours of ICU stay are used. Insulin sensitivity (SI) is identified hourly, as well as the hour-to-hour percentage change in SI (%ΔSI). Differences between males and females SI and %ΔSI over 6-h blocks are compared using hypothesis and equivalence testing. A difference in SI levels would suggest a difference in metabolic stress response to insult, while a difference in %ΔSI levels would suggest a resulting difference in the difficulty to control.Results show females are significantly more insulin resistant than males and not equivalent, suggesting stronger stress response to insult induced stress. Metabolic variability is equivalent in both groups, advocating GC safety and efficacy should be similar between males and females, despite potential higher insulin rates required for females.This study is the first to suggest potential gender differences in the metabolic stress response.

Model-free adaptive control for MEA-based post-combustion carbon capture processes

For the flexible operation of mono-ethanol-amine-based post-combustion carbon capture processes, recent studies concentrate on model-based protocols which require underline model parameters of carbon capture processes for controller design. In this paper, a novel application of the model-free adaptive control algorithm is proposed that only uses measured input-output data for carbon capture processes. Compared with proportional-integral control, the stability of the closed-loop system can be easily guaranteed by increasing a stabilizing parameter. By updating the pseudo-partial derivative vector to estimate a dynamic model of the controlled plant on-line, this new protocol is robust to plant uncertainties. Compared with model predictive control, tuning tests of the protocol can be conducted on-line without non-trivial repetitive off-line sensitivity or identification tests. Performances of the model-free adaptive control are demonstrated within a neural network carbon capture plant model, identified and validated with data generated by a first-principle carbon capture model.

Modelling Intestinal Glucose Absorption using Continuous Glucose Monitor Data

Hyperglycaemia is a common complication of prematurity in neonatal intensive care. Physiological model-based protocols offer great potential for allowing patient-specific insulin therapy. One aspect of such models is the appearance of glucose in the bloodstream from enteral feeds. A two compartment model is used to capture glucose transport in the stomach and gut. The rate of transfer of glucose from the stomach compartment to the gut can be determined from gastric emptying, which in term infants can be modelled with a half life of around 20 minutes (d1=0.035 min−1). In premature infants gastric emptying is slower, so can be modelled with a half life of around 40 minutes (d1=0.017 min−1). CGM data showing the appearance of glucose from enteral feeds in term and premature infants was used to estimate the rate constant for glucose absorption from the gut to the bloodstream. This absorption rate constant was found to vary between infants and over time. The median absorption constant was found to be d2=0.014 min−1, which corresponds to an absorption half life of 50 minutes. If fitting error that fell within blood glucose measurement error was considered, the range of possible absorption half lives of absorption was 30-63 minutes.

Costs and Benefits of Implementing Sensor-controlled Irrigation in a Commercial Pot-in-Pot Container Nursery

Irrigation scheduling in ornamental plant production is complex due to the large number of species grown by individual growers and the need to consider plant, environment, and substrate conditions to make correct irrigation decisions on a daily or more frequent basis. The engineering team in our project has developed a smart wireless sensor node that is capable of integrating outputs from a range of soil moisture and environmental sensors to schedule irrigation events. In addition, an advanced monitoring and control software enables growers to manage irrigation based on set-point or model-based protocols, which are then independently executed by the nodes, enhancing or replacing human decision making. During 2012, we implemented a sensor-controlled vs. grower-controlled irrigation study at a pot-in-pot nursery in Tennessee. Sensor networks were installed in two separate production blocks of 3-year-old dogwood (Cornus florida ‘Cherokee Brave’) and 2-year-old red maple (Acer rubrum ‘Autumn Blaze’) trees grown in 15- and 30-gal containers, respectively. One row of trees in each block was irrigated based on the average reading of soil moisture sensors inserted in individual trees using micropulse irrigation, i.e., sensor controlled. Trees in the adjacent row and the rest of the block were independently irrigated by the grower using standard practices, i.e., grower controlled. Sensor volumetric water content (VWC) readings and irrigation volumes were logged by nodes on a 15-min basis and were relayed to a base station on the farm. For the study period between Mar. 2012 and Nov. 2012, average daily water applied by the grower-controlled irrigation to the dogwood block was 0.92 gal/tree, compared with 0.34 gal/tree applied using sensor-controlled irrigation; for red maple, these values were 1.72 gal/tree and 1.13 gal/tree, respectively. No significant differences in tree caliper or quality were noted between the two irrigation treatments in either species over the year. The cost of water for this particular operation was negligible consisting only of pumping costs, as water is drawn from a perennial stream with excellent water quality. Consequently, a conservative return on investment for a wireless sensor network capable of covering the entire operation was 37.5%, corresponding to a payback period of 2.7 years, associated almost entirely from a reduction in irrigation management time. Pricing in a nominal cost for water of $326 per acre-foot ($1 per 1000 gal) increased annualized net savings 9-fold, reducing the payback period to less than 4 months. This analysis did not factor in additional economic benefits such as reductions in production time, losses due to disease, or increased plant quality, which have been associated with the use of sensor-based irrigation control in other studies.

Privacy-Preserving Speaker Verification and Identification Using Gaussian Mixture Models

Speech being a unique characteristic of an individual is widely used in speaker verification and speaker identification tasks in applications such as authentication and surveillance respectively. In this article, we present frameworks for privacy-preserving speaker verification and speaker identification systems, where the system is able to perform the necessary operations without being able to observe the speech input provided by the user. In a speech-based authentication setting, this privacy constraint protect against an adversary who can break into the system and use the speech models to impersonate legitimate users. In surveillance applications, we require the system to first identify if the speech recording belongs to a suspect while preserving the privacy constraints. This prevents the system from listening in on conversations of innocent individuals. In this paper we formalize the privacy criteria for the speaker verification and speaker identification problems and construct Gaussian mixture model-based protocols. We also report experiments with a prototype implementation of the protocols on a standardized dataset for execution time and accuracy.

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 5600–144 000 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%.

Emergency Medical Services for Children: Children and their Accidents

<h3>Objective</h3> Previous studies examine clinical outcomes of insulin therapy in neonatal intensive care units (NICUs), without first developing safe and effective control protocols. This research quantifies the safety and performance of a computerised model-based control algorithmSTAR-GRYPHON (Stochastic TARgeted Glucose Regulation sYstem to Prevent Hyper- and hypO-glycaemia in Neonates). <h3>Design</h3> Retrospective observational study of glycaemic control in very/extremely low birthweight infants treated with insulin from Christchurch Women’s Hospital NICU between January 2013 and June 2017. Blood glucose (BG) outcomes and control performance is compared with retrospective data (n=22) and literature. <h3>Interventions</h3> Insulin infusion doses were calculated from 3 to 4 hourly BG measurements using a computerised model-based control algorithm, STAR-GRYPHON. <h3>Main outcome measures</h3> Mean BG, time in targeted range and incidence of hypoglycaemia. <h3>Results</h3> STAR-GRYPHON (n=35) had lower mean BG concentration (7.0mmol/L vs 7.9 mmol/L), higher %BG within the 4.0–8.0 mmol/L target range (71.1% vs 50.9%) and lower %BG &lt;4.0 mmol/L (0.6% vs 2.1%). There were only 2 BG &lt;2.6 mmol/L (over n=2, 5.5% of patients, 0.03% of all BG outcomes), one of which may be attributed to clinical error. These results show better control to target and lower incidence of hypoglycaemia than most literature results from intensive insulin therapy protocols or study groups in children and infants. <h3>Conclusions</h3> Model-based protocols can safely and effectively control BG in very premature infants and should be used in future studies to determine the effect of insulin therapy on clinical outcomes.