PREVENTING VENTILATOR-ASSOCIATED LUNG INJURY USING FEEDBACK CONTROL ENGINEERING

本项目将建立肺泡通气动力学（包括气体交换、动态和非线性肺泡顺应性和支气管阻力）、心血管性能和血液（参考其携气能力）的多室、多标量数学模型。所建立的模型将用于阐明在病变的、异质的、机械通气的肺中引起肺损伤的因素的程度和分布。通过将该问题视为反馈控制之一，我们将研究机械呼吸机的参数调整方法，以优化心输出量和动脉气体张力，同时最小化与VALI相关的因素。由于我们打算开发的仿真模型的不可避免的复杂性，将需要来自多变量鲁棒和最优控制理论的先进方法，以确定应该调整哪些参数组合，以及如何调整，以实现所需的VALI减少。这项工作将超越之前的尝试，通过提出的模拟平台的更高保真度来量化机械通气期间肺损伤风险因素。此外，我们将鲁棒性分析技术应用于建模，以提高我们研究结果的可靠性和适用性。这将使我们能够进行人口建模，而不是通常使用的建模和研究单一，理想化主题的方法，使我们的发现适用于人群和各种真实患者，与之前的工作相反，理想化的主题实际上既不代表人群也不代表任何个体。

This project will develop a multi-compartmental, multi-scalar, mathematical model of alveolar ventilation dynamics (which includes gas exchange, dynamic and non-linear alveolar compliance and bronchial resistance), cardiovascular performance and blood (with reference to its gas-carrying abilities). The developed model will be used to elucidate the extent and distribution of the factors causative of lung injury in diseased, heterogeneous, mechanically-ventilated lungs. By treating the problem as one of feedback control, we will investigate methods of parameter adjustment in the mechanical ventilator to optimise cardiac output and arterial gas tensions while minimising the factors associated with VALI. Due to the inevitable complexity of the simulation model which we intend to develop, advanced methods from multivariable robust and optimal control theory will be required in order to identify which combinations of parameters should be adjusted, and how, in order to achieve the desired reduction in VALI. The work will go beyond that previously attempted in quantifying the factors that risk lung injury during mechanical ventilation through the greater fidelity of the proposed simulation platform. In addition, we will apply robustness analysis techniques to the modelling to improve the reliability and applicability of our findings. This will allow us to perform population modelling, rather than the commonly used approach of modelling and studying a single, idealized subject, rendering our findings applicable to populations and to a variety of real patients, in contrast to previous work where the idealized subject is in fact representative of neither the population nor any one individual.