老年患者大手术术后谵妄的脑功能特征与术前预测模型研究

Postoperative delirium (POD) has been recognized as the most prevalent post-surgical complication among the elder people and often leads to prolonged stay in the hospital, long-term cognitive decline, and even mortalities. The occurrence of POD can be reduced considerably by providing a preoperative intervention to the high-risk individuals. Identification of the high-risk patient however remains challenging mainly due to the lack of understanding about the pre-disposing factor of POD. This scenario gets even worse when the brain related risk factors exist and remain unclear. As such, in this proposal, our prime focus is at deciphering the brain related risk factor to POD. During the course of this project, we are planning to design a cognitive and resting-state experiment to identify the neural basis of the well-known cognitive abnormalities observed in POD patients as the first step. Simultaneously acquired EEG and fNIRS recordings during the experiment will be employed to identify the brain region related to high POD risk by performing the multimodal imaging fusion. We hypothesize that the estimated connectivity strength among the high-risk related brain regions and the whole-brain functional network can be employed to identify the risk factor of POD from a brain functional perspective. Our ultimate goal of this research is to realize a unified POD risk prediction model by fusing the features obtained through the brain network analysis, the behavior test, and the evoked related brain responses. We further hypothesize that the proposed research project could provide new evidence in understanding the neural basis for POD. Furthermore, it can provide theory basis and necessary methods for the clinicians to enhance the reliability while performing POD high-risk screening protocol.

术后谵妄是一种严重威胁老年手术患者生存质量的常见术后并发症。通过对患者进行术前风险筛查，针对高危人群实施对症干预可以有效降低术后谵妄的发生率。然而，由于术后谵妄易感因素的复杂性，特别是术后谵妄患者的脑功能风险特征仍不明确，致使现有术前预测模型的有效性无法得到保证。由此，本项目基于认知储备和神经系统衰老理论，结合术后谵妄患者的特点，设计针对工作记忆和注意力的任务态及静息态实验，同步采集EEG-fNIRS神经信号，通过多模态信息融合，确定术后谵妄患者的高风险脑区，结合高风险脑区间的连接强度和全脑功能网络分析，重点探索术后谵妄患者术前脑功能的特征。最终构建融合行为学测试、刺激诱发脑活动、术前一般风险因素和脑功能网络特征的术后谵妄预测新模型。本项目的实施可为理解术后谵妄发病的神经机制提供新的证据支持，为临床上有效进行术后谵妄高危患者筛查提供理论和工具支持。

本项目针对老年患者术后谵妄与术后认知功能异常高发这一临床问题，采集患者术前和术中的EEG信号，借助MMN实验范式，通过EEG信号的时域、频域、脑功能网络连接强度、脑功能网络测度等特征，识别术后谵妄高危患者的EEG特征，借助机器学习技术构建术前和术中的老年谵妄患者筛选模型。同时，我们针对麻醉药物丙泊酚的脑功能网络变化机制以及药物个体差异性，开展了基于脑功能网络和EEG微状态的丙泊酚麻醉药物作用神经机制研究。主要完成了以下几点工作：.1）形成具有完备临床资料的老年患者术前与书中EEG数据库；2）确定了老年患者术后谵妄与术后认知功能异常的术中与术前脑电特征，构建了通过术中脑电特征的术后谵妄与术后认知功能异常预测分析模型；3）系统性研究了麻醉药物丙泊酚致无意识过程中的大脑功能网络的变化过程，确定丙泊酚的主要作用脑区以及丙泊酚对动态和静态脑功能网络作用方式；4）通过麻醉前动态脑功能网络特征，探索了丙泊酚个体响应差异的术 前脑功能网络特征；。.本项目的主要发现有：.1）较小的频率差所诱导的MMN波形幅值差异与患者术后谵妄概率显著相关；2）在对脑电信号进行高维度特征提取，并辅以LASSO特征筛选后，我们可以获得准确率为82%的术前谵妄风险评估模型；3）通过对术中额区EEG信号进行特征分解，我们识别出麻醉过程中的慢波占比与术后谵妄风险显著相关；4）更高的动态脑功能网络强度与个体对丙泊酚的响应密切相关。我们的结果为更好的对患者进行麻醉药物输注提供了重要理论支持；5）随着麻醉深度的加深，大脑高级到低级皮层之间的连接值显著下降，但是这种下降在浅麻醉时显示出饱和的状态。由此我们发现额顶功能连接与麻醉水平相关，然而无法作为有效的麻醉深度监测指标；6）发现了麻醉状态下会产生有别于清醒状态下的新的脑微状态，使用微状态的特征，可以实现高精度的意识水平分类。