Collaborative Research: Multivariable Modeling and Control of Clinical Pharmacodynamics

智力优势：在手术过程中，麻醉师不断调整给予患者的麻醉剂输送，以保持足够的麻醉深度。同时，麻醉师维持通气参数并监测心血管和呼吸功能。根据麻醉师的专业知识结合他或她对患者输出（如血压、心率、呼出气体和基于EEG的测量）的观察，确定患者何时被正确麻醉。 也就是说，麻醉师在手术过程中扮演多变量反馈控制器的角色。麻醉剂给药的自动化将允许最佳定制给予患者的麻醉剂量，并允许麻醉师专注于预期和意外的患者手术需求所必需的关键任务。实施闭环药物递送的直接优势将是降低药物成本，缩短恢复时间和改善患者的长期结果。为了设计反馈控制方案，需要患者/药物输送系统的数学模型，其适合于控制目的。在本研究项目中，我们集中于（1）开发与控制相关的多输入、多输出模型，以描述患者对麻醉剂、通气控制和外部刺激的反应，以及（2）控制策略的开发和实施，从而改善患者安全性和术后结局。我们的目标是系统的先进多变量控制和识别技术的发展，我们需要充分解决这个问题。具体工作将包括开发控制合成、多变量切换线性系统的分析和系统识别算法，以及直接针对多变量分室系统经验建模的模型结构规范和子空间系统识别算法。手术室中越来越多地使用计算机，再加上最近非手术室的发展，测量麻醉的许多目标的侵入性但有效的方法承诺将控制技术结合到麻醉输送中过程迫在眉睫。该项目的结果将对临床药理学中的主动控制应用的状态产生巨大而显著的影响，尤其是对手术期间麻醉深度的临床监测和控制。PI已与General Electric Healthcare和ASPECT Medical Systems，Inc.的临床监测和围手术期研发小组建立了联系，通过这一机制，确保技术转让和与工业实施的相关性。

Intellectual Merit:During surgery, anesthesiologists continuously adjust the delivery of anesthetic agents given to the patient in order to maintain an adequate level of anesthetic depth. Simultaneously, anesthesiologists maintain ventilation parameters and monitor cardiovascular and respiratory functions. The determination of when a patient is properly anesthetized is made based on the anesthesiologist's expertise combined with his or her observations of patient outputs such as blood pressure, heart rate, exhaled gases, and EEG-based measures. That is, anesthesiologists perform the role of multivariable feedback controllers during surgery. Automating the administration of anesthetics will allow both for optimal tailoring of the amount of anesthetics given to patients and for the anesthesiologist to focus on critical tasks necessitated by surgical demands on the patient that are both expected and unexpected. The direct advantages of implementing closed-loop drug delivery would be reduced pharmaceutical costs, reduced recovery time and improved long-term patient outcomes. In order to design feedback control schemes, mathematical models of the patient/drug delivery system that are suitable for control purposes are required.In this research project we focus on (1) the development of control-relevant multi-input, multi-output models to describe patient response to anesthetic agents, ventilation controls and external stimuli, and (2) the development and implementation of control strategies for which patient safety and postoperative outcomes are improved. We target the development of systematic advanced multivariable control and identification techniques, which we posit are required to adequately address this problem. Specific efforts will include the development of control synthesis, analysis and system identification algorithms for multivariable switched-linear systems, and model structure specification and subspace system identification algorithms directly aimed at empirical modeling of multivariable compartmental systems.Broader Impact:The increasing use of computers in the operating room combined with the recent development of non-invasive yet effective means of measuring a number of the goals of anesthesia promise to make the incorporation of control techniques into the anesthetic delivery process imminent. The results of this project will have a dramatic and significant impact on the state of active control applications in clinical pharmacology, in general, and clinical monitoring and control of anesthetic depth during surgery, more specifically. The PIs have established connections with clinical monitoring and perioperative research and development groups at General Electric Healthcare, and ASPECT Medical Systems, Inc., through which technology transfer and relevance to industry implementation is assured.