CAREER: Enhancing perception and cognition while minimizing side effects through closed-loop peripheral neural stimulation

知觉、认知和行为表现受唤醒水平的影响很大。例如，学生在清醒时的考试成绩可能会比困倦或过度焦虑时好得多。觉醒是由几个神经调节系统调节的，包括蓝斑（LC），这是脑干中的一群神经元，其损伤经常导致神经退行性疾病。因此，从工程学的角度来看，控制LC活性可以最大限度地提高健康个体的行为表现并治疗LC相关疾病是合理的。然而，LC在脑干中的小尺寸和深位置使其难以进入和操作。为了克服这一挑战，本项目将使用外周迷走神经刺激（VNS）来控制LC活动。迷走神经从脑干延伸至结肠区域，易于无创接近，LC是介导VNS对脑活动影响的主要脑结构。利用尖端的机器学习和非线性控制理论，本项目旨在开发一个工程框架，通过外围VNS和测量瞳孔大小来控制LC活动，瞳孔大小也是唤醒水平的一个指标。该项目的成功将导致非侵入性方法的发展，以增强人类的感知和认知，并治疗与lc相关的脑部疾病，如阿尔茨海默病、双相情感障碍、抑郁症和注意缺陷多动障碍（ADHD）。计划中的研究活动将丰富本科课程，为本科生和研究生提供跨学科的研究机会，为快速发展的生物技术和生物医学工程领域的国家高素质劳动力做出贡献。此外，该项目有望扩大代表性不足的少数民族和社会经济背景不利的女学生对STEM的参与。拟议的教育课程和外展活动也将吸引公众，提高他们对如何应用STEM学科来解决神经系统问题的认识。PI的长期职业研究目标是通过控制神经系统内的人口活动来增强人类的感知和认知。为了实现这一目标，本项目将开发一种工程框架，该框架使用周围神经刺激来控制神经群活动，以实现最佳行为表现，同时最大限度地减少意外副作用，并在动物模型中对该技术进行功能验证。迷走神经到蓝斑（LC）通路将作为模型系统来开发和测试该技术。之所以选择这一途径，是因为FDA批准了一种非侵入性迷走神经刺激器的存在，并且LC通过调节觉醒水平在调节大脑功能中起着关键作用。研究将包括LC尖峰活动和瞳孔大小的评估，这是LC活动的无创代理，将在未来的技术翻译中有用。研究计划有三个目标。第一个目标是确定VNS（迷走神经刺激）的最佳参数空间，以选择性地调节LC活动/瞳孔大小，同时使用贝叶斯主动学习最小化对心率的副作用。一组动物将被植入颈动脉水平的左VNS电极（钝性剥离）和无线心电图（ECG）发射器来监测心率。一个基因编码的钙指示剂将在LC神经元中选择性地表达，利用病毒载体，实现LC神经元群体活动的钙成像。LC神经元的平均尖峰活动将从荧光信号中推断出来。贝叶斯主动学习将用于发现VNS参数空间，其中VNS在驱动LC活动/瞳孔大小方面效率高，同时对心率的影响最小。刺激将以短脉冲间隔的电荷平衡双相电流脉冲爆发的形式呈现。参数空间将包括脉冲高幅相的电流幅值、高幅相的脉冲宽度、脉冲极性（即阴极领先或阳极领先）、脉冲波形、不对称比、脉冲间隔、每次突发脉冲数和突发间隔。第二个目标是合成一个非线性闭环控制器，用于控制LC群体活动/瞳孔大小，同时通过VNS最小化心率变化。非参数高斯过程（GP）模型将用于模拟迷走神经到lc和迷走神经到瞳孔回路的非线性动力学。基于GP模型合成非线性预测模型控制器，通过VNS以闭环方式控制LC种群活动/瞳孔大小。不同的非线性优化方法将在控制器中进行测试，并进行大量的仿真来评估其性能。第三个目标是在清醒行为的动物身上对该技术进行功能验证。控制器在多大程度上能够保持最佳的行为表现，同时在实际应用中对动物产生最小的副作用，将进行研究。用于监测LC神经元尖峰表现的大鼠将接受训练，以执行触觉检测任务，例如，舔水管以响应须偏转。他们的行为表现将从他们对感官刺激的正确和错误反应的概率来定量评估。将打开刺激器的动物的行为表现和心率与关闭刺激器的动物进行比较。将测试不同的射击速率，以确定哪种射击速率能产生最佳的行为性能。上述目标的成功完成将首次提供一个工程框架，用于指导优化闭环增产措施的设计和验证。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Perception, cognition, and behavioral performance are heavily influenced by arousal level. For example, a student will likely perform much better on exams when alert than when drowsy or overly anxious. Arousal is regulated by several neuromodulatory systems, including the locus coeruleus (LC), which is a cluster of neurons in the brainstem whose damage often leads to neurodegenerative diseases. Thus, from an engineering standpoint, it is plausible that control of LC activity could maximize behavioral performance in healthy individuals and treat LC related disorders. However, the small size and deep location of the LC in the brainstem makes it challenging to access and manipulate. To overcome this challenge, this project will use peripheral vagus nerve stimulation (VNS) to control LC activity. The vagus nerve, which runs from the brainstem to the colon area, is easy to access non-invasively, and the LC is the main brain structure that mediates the effect of VNS on brain activity. Using cutting-edge machine learning and nonlinear control theory, this project aims to develop an engineering framework for controlling LC activity via periphery VNS and measuring pupil size, which is also an indicator of arousal level. The success of this project will lead to the development of non-invasive methods to enhance human perception and cognition and treat LC-related brain disorders, e.g., Alzheimer's disease, bipolar disorder, depression and Attention Deficit Hyperactivity Disorder (ADHD). Research activities in the planned studies will enrich the undergraduate curriculum and provide interdisciplinary research opportunities for undergraduates and graduate students, who will contribute to the national highly-qualified workforce in the rapidly-growing biotechnology and biomedical engineering field. Moreover, the project is expected to broaden the participation in STEM of under-represented minority and female students with social-economically disadvantaged backgrounds. The proposed educational curriculum and outreach activities will also engage the general public to increase their awareness of how STEM disciplines can be applied to solve problems in the nervous system.The PI's long-term career research goal is to enhance human perception and cognition through control of population activities within the nervous system. Toward this goal, this project is to develop an engineering framework that uses peripheral nerve stimulation to control neural population activity to achieve optimal behavioral performance while minimizing unintended side effects and to functionally validate this technology in an animal model. The vagus nerve-to-locus coeruleus (LC) pathway will be used as a model system to develop and test the technology. This pathway was chosen because an FDA approved non-invasive vagus nerve stimulator exists and the LC plays a pivotal role in modulating brain functions through regulation of arousal levels. Studies will include assessment of LC spiking activity and pupil size, which is a noninvasive proxy of LC activity that will be useful in future translation of the technology developed. The Research Plan is organized under three objectives. The FIRST OBJECTIVE is to identify the optimal parameter space of VNS (vagus nerve stimulation) for selectively modulating LC activity/pupil size while minimizing side effects on heart rate using Bayesian active learning. A cohort of animals will be implanted with electrodes for left VNS at the level of the carotid artery (blunt dissection) and with a wireless electrocardiography (ECG) transmitter to monitor heart rate. A genetically-encoded calcium indicator will be selectively expressed in LC neurons using viral vectors, enabling calcium imaging of the activity of LC populations of neurons. The average spiking activity of the LC neurons will then be inferred from the fluorescence signals. Bayesian active learning will be used to discover the VNS parameter spaces in which VNS has high efficacy in driving LC activity/pupil size while leaving minimal effects on heart rate. Stimulation will be presented in the form of bursts of charge-balanced bi-phasic current pulses with short inter-pulse intervals. The parameter space will include current amplitude of the higher amplitude phase of the pulses, pulse width of the higher amplitude phase, pulse polarity (i.e. cathode-leading or anode-leading), pulse waveform, asymmetry ratio, inter-pulse interval, number of pulses per burst, and inter-burst interval. The SECOND OBJECTIVE is to synthesize a nonlinear closed-loop controller for control of LC population activity/pupil size while minimizing changes in heart rate through VNS. Non-parametric Gaussian Process (GP) models will be used to model the nonlinear synamics of the vagus nerve-to-LC and vagus nerve-to-pupil circuitry. A nonlinear predictive model controller based on the GP models will be synthesized to control LC population activity/pupil size through VNS in a closed-loop fashion. Different nonlinear optimization methods will be tested in the controller, and extensive simulations will be conducted to evaluate its performance. The THIRD OBJECTIVE is to functionally validate the technology in awake behaving animals. The extent to which the controller is able to maintain optimal behavioral performance while producing minimal side effects for animals in real-world applications will be examined. Rats instrumented for monitoring the spiking performance of neurons in the LC will be trained to perform a tactile detection task, e.g., licking a water spout in response to a whisker deflection. Their behavioral performance will be quantitatively evaluated from the probabilities of their correct and incorrect responses to sensory stimuli. The behavioral performance and heart rate of animals in sessions with the stimulator on will be compared to those in sessions with the stimulator off. Different firing rates will be tested to determine which firing rate results in optimal behavior performance. The successful completion of the above objectives will provide, for the first time, an engineering framework to guide the design and validation of optimal, closed-loop stimulation for enhancing behavior.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.