Dynamically generated models for medical decision support systems

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.

Computerized Decision Support Systems for Mechanical Ventilation in Children

Mechanical ventilation is an effective treatment in the ICU but can have significant adverse effects. Approaches from adult research have been adopted in pediatric critical care despite known differences in respiratory physiology and ICU processes. There continues to be considerable variation in how ventilators are managed. Computerized decision support systems implement explicit protocols, and are designed to make mechanical ventilation management safer, more consistent, and more lung protective. Variable results and low or unknown compliance with protocols and CDSS tools have been reported. To date, there has been limited research regarding CDSS for mechanical ventilation in children.

A java based simulator with user interface to simulate ventilated patients

Mechanical ventilation is a life-saving intervention, which despite its use on a routine basis, poses the risk of inflicting further damage to the lung tissue if ventilator settings are chosen inappropriately. Medical decision support systems may help to prevent such injuries while providing the optimal settings to reach a defined clinical goal. In order to develop and verify decision support algorithms, a test bench simulating a patient’s behaviour is needed. We propose a Java based system that allows simulation of respiratory mechanics, gas exchange and cardiovascular dynamics of a mechanically ventilated patient. The implemented models are allowed to interact and are interchangeable enabling the simulation of various clinical scenarios. Model simulations are running in real-time and show physiologically plausible results.

Simulating physiological interactions in a hybrid system of mathematical models

Mathematical models can be deployed to simulate physiological processes of the human organism. Exploiting these simulations, reactions of a patient to changes in the therapy regime can be predicted. Based on these predictions, medical decision support systems (MDSS) can help in optimizing medical therapy. An MDSS designed to support mechanical ventilation in critically ill patients should not only consider respiratory mechanics but should also consider other systems of the human organism such as gas exchange or blood circulation. A specially designed framework allows combining three model families (respiratory mechanics, cardiovascular dynamics and gas exchange) to predict the outcome of a therapy setting. Elements of the three model families are dynamically combined to form a complex model system with interacting submodels. Tests revealed that complex model combinations are not computationally feasible. In most patients, cardiovascular physiology could be simulated by simplified models decreasing computational costs. Thus, a simplified cardiovascular model that is able to reproduce basic physiological behavior is introduced. This model purely consists of difference equations and does not require special algorithms to be solved numerically. The model is based on a beat-to-beat model which has been extended to react to intrathoracic pressure levels that are present during mechanical ventilation. The introduced reaction to intrathoracic pressure levels as found during mechanical ventilation has been tuned to mimic the behavior of a complex 19-compartment model. Tests revealed that the model is able to represent general system behavior comparable to the 19-compartment model closely. Blood pressures were calculated with a maximum deviation of 1.8 % in systolic pressure and 3.5 % in diastolic pressure, leading to a simulation error of 0.3 % in cardiac output. The gas exchange submodel being reactive to changes in cardiac output showed a resulting deviation of less than 0.1 %. Therefore, the proposed model is usable in combinations where cardiovascular simulation does not have to be detailed. Computing costs have been decreased dramatically by a factor 186 compared to a model combination employing the 19-compartment model.

Efficient computation of interacting model systems

Physiological processes in the human body can be predicted by mathematical models. Medical Decision Support Systems (MDSS) might exploit these predictions when optimizing therapy settings. In critically ill patients depending on mechanical ventilation, these predictions should also consider other organ systems of the human body. In a previously presented framework we combine elements of three model families: respiratory mechanics, cardiovascular dynamics and gas exchange. Computing combinations of moderately complex submodels showed to be computationally costly thus limiting the applicability of those model combinations in an MDSS. A decoupled computing approach was therefore developed, which enables individual evaluation of every submodel. Direct model interaction is not possible in separate calculations. Therefore, interface signals need to be substituted by estimates. These estimates are iteratively improved by increasing model detail in every iteration exploiting the hierarchical structure of the implemented model families. Simulation error converged to a minimum after three iterations. Maximum simulation error showed to be 1.44% compared to the original common coupled computing approach. Simulation error was found to be below measurement noise generally found in clinical data. Simulation time was reduced by factor 34 using one iteration and factor 13 using three iterations. Following the proposed calculation scheme moderately complex model combinations seem to be applicable for model based decision support.

Structural identifiability and practical applicability of an alveolar recruitment model for ARDS patients

Patient-specific mathematical models of respiratory mechanics can offer substantial insight into patient state and pulmonary dynamics that are not directly measurable. Thus, they offer significant potential to evaluate and guide patient-specific lung protective ventilator strategies for acute respiratory distress syndrome (ARDS) patients. To assure bedside applicability, the model must be computationally efficient and identifiable from the limited available data, while also capturing dominant dynamics and trends observed in ARDS patients. In this study, an existing static recruitment model is enhanced by considering alveolar distension and implemented in a novel time-continuous dynamic respiratory mechanics model. The model was tested for structural identifiability and a hierarchical gradient descent approach was used to fit the model to low-flow test responses of 12 ARDS patients. Finally, a comprehensive practical identifiability analysis was performed to evaluate the impact of data quality on the model parameters. Identified parameter values were physiologically plausible and very accurately reproduced the measured pressure responses. Structural identifiability of the model was proven, but practical identifiability analysis of the results showed a lack of convexity on the error surface indicating that successful parameter identification is currently not assured in all test sets. Overall, the model presented is physiologically and clinically relevant, captures ARDS dynamics, and uses clinically descriptive parameters. The patient-specific models show the ability to capture pulmonary dynamics directly relevant to patient condition and clinical guidance. These characteristics currently cannot be directly measured or established without such a validated model.

Sequential versus concurrent computation of complex model systems for medical decision support

Medical Decision Support Systems employ mathematical models to optimize therapy settings. The mathematical models are used to predict patient reactions towards alteration in the therapy regime. This prediction should not be limited to one detail but should feature a broad picture. A previously proposed framework is able to dynamically combine submodels of three model families (respiratory mechanics, gas exchange and cardiovascular dynamics) to form a complex, interacting model system. When concurrent computation of the combined submodels is employed, tests exhibited high computing costs. Therefore, a sequential computing approach is introduced. Thereby, direct interaction between the submodels is not applicable as all submodels are computed individually. To simulate submodel interaction, interface signals that are normally present in the concurrent approach were precalculated using reduced models of respiratory mechanics and cardiovascular dynamics. Evaluation of the new approach showed that results feature a discrepancy lower than 2.5% compared to the results computed by the concurrent approach. Simulation error could be decreased to 2% by improving the precalculation of the interface signals. Computing costs have been decreased by a factor of 17.