Learning to Predict Delayed Cerebral Ischemia with Novel Continuous Cerebral Arterial State Index

项目摘要 迟发性脑缺血是动脉瘤术后最具破坏性的并发症。 蛛网膜下腔出血(ASAH)，发病率为30%。目前的做法依赖于 间歇性评估神经功能状态和每日脑血流速度(CBFV) 经颅多普勒超声(TCD)用于指导医疗管理，预防DCI。仅在之后 医疗管理失败，血管内治疗(EVT)包括动脉内血管扩张剂输注 和/或颅内血管成形术开始。这种被动的做法不能解释早期的预测因素 DCI并可能在出现症状或症状之前错过DCI发展的早期阶段的最佳EVT窗口 严重偏离正常血流动力学。这个项目的目标是开发算法来 在发展的早期阶段预测DCI和相关目标。对DCI的准确预测 将能够采取更积极主动的战略来预防和治疗DCI的根本原因。 为实现该项目的目标，将实现以下三个目标：1)开发特定于ASAH的 基于颅内压脉搏的脑动脉状态指数；2)开发和验证预测 与ASAH后迟发性脑缺血相关的靶点模型；3)进行前瞻性机构- 对所开发的模型进行具体的改编和验证。 我们的DCI预测算法只需要当前临床实践中可用的数据，因此它们可以 很容易被采用。如果得到验证，这些算法将使临床医生能够监控DCI的风险 持续和主动地提供适当的治疗。建议进行的前瞻性研究 算法的实现和适配将为未来的临床试验做好准备，以测试 算法知情的干预。

Project Summary Delayed cerebral ischemia (DCI) is the most devastating complication after aneurysmal subarachnoid hemorrhage (aSAH) and has an incidence rate of 30%. Current practice relies on intermittent assessment of neurological status and daily cerebral blood flow velocity (CBFV) by Transcranial Doppler ultrasound (TCD) to guide medical management to prevent DCI. Only after medical management fails, is endovascular treatment (EVT) including intraarterial vasodilator infusion and/or intracranial angioplasty initiated. This reactive practice does not account for early predictors of DCI and may miss the optimal EVT window at an early stage of DCI development before symptoms or severe deviations from normal hemodynamics. The goal of this project is to develop algorithms to predict DCI and related targets at an early stage in their development. An accurate prediction of DCI will enable a more proactive strategy to prevent and treat the underlying cause of DCI. The following three aims will be pursued towards the goal of the project: 1) Develop aSAH-specific intracranial pressure (ICP) pulse-based cerebral arterial state index; 2) Develop and validate predictive models of targets related to delayed cerebral ischemia after aSAH; 3) Conduct a prospective institution- specific adaption and validation of the developed models. Our DCI predictive algorithms only need data available in current clinical practice hence they can be readily adopted. If validated, these algorithms will enable clinicians to monitor risk of DCI continuously and to proactively deliver appropriate treatment. The proposed prospective study of algorithm implementation and adaptation will well prepare future clinical trials to test the efficacy of algorithm-informed interventions.