Bio-digital Rapid Alert to Identify Neuromorbidity

在住院患者中， 重症患者代表了一种新认识和新出现的流行病。其中包括入院的患者， 重症监护病房与主要神经系统诊断，那些在增加风险的基础上，他们的基础 疾病（例如嗜神经病毒感染，包括COVID 19），以及那些 神经疾病是隐匿的和不可预料的。与危重病相关的神经发病可能由以下原因引起： 生理不稳定、生化紊乱、药物副作用、侵入性机械支持， 不动和/或用于防止死亡的其他疗法。它跨越了从新生儿到 老年人，发生于不同性别和种族，并且在全身性疾病（例如败血症， 病毒感染和其他炎性疾病）和严重器官衰竭（如急性呼吸窘迫 综合征、癌症、肝和肾衰竭）。在美国，神经病的发病率范围为5-47%， 重症儿童和成人，因此每年影响数十万患者。经常 神经病的发展直到临床表现出现后才被发现，并且是不可逆的。神经病态 可急性发作，例如癫痫发作、中风、脑内出血、脑水肿和/或谵妄，或 更持久的方式，例如神经肌肉无力和/或认知能力下降，并且通常是永久性的， 在一个人的余生中忍受。没有标准的临床工具来识别 神经发病或实时神经监测的风险，与心脏、肾脏、肝脏和 许多其他器官。 为了填补这一空白，改变临床医生检测和监测脑损伤进展的方式，我们开发了 一个生物数字快速警报，以确定神经疾病（大脑），不断饲料电子健康记录 (EHR)将9个临床领域（A至I）的变量导入专有信息学和机器学习平台。 原型BRAIN A-I模型是稳健的，并在临床行动之前预测临床医生对神经发病率的关注。 满员你为了将生物和数字签名联系起来，我们定义了一组血清生物标志物， 在临床检测之前有时间记录的神经发病率。使用“贝叶斯到床边”方法，我们有 创建了一个连接EHR和专用主机服务器的实时数据管道， 这是将BRAIN A-I作为嵌入式预测分析和决策驱动支持工具进行操作所必需的。 在这项提案中，我们将测试EHR中的数字签名与大脑特异性 生物标志物可以快速检测危重儿童的神经发病率。成功部署可互操作、 24/7床旁神经监测用于早期发现神经发病率将是一个突破 用于重症患者的临床管理。

The silent development and progression of neurologic morbidity, or neuromorbidity, among hospitalized, critically ill patients represents a newly recognized and emerging epidemic. This includes patients admitted to intensive care units with primary neurologic diagnoses, those at increased risk based on their underlying disease (e.g. neurotropic viral infections including COVID19), and those where the development of neuromorbidity is occult and unexpected. Neuromorbidity associated with critical illness can be caused by physiologic instability, biochemical derangements, side effects of medications, invasive mechanical support, immobility, and/or other therapies used to prevent death. It spans the age spectrum from neonates to the elderly, occurs across gender and race, and is underrecognized in patients with systemic illnesses (e.g. sepsis, viral infections, and other inflammatory conditions) and critical organ failure (e.g. acute respiratory distress syndrome, cancer, hepatic and renal failure). In the U.S. the incidence of neuromorbidity ranges from 5-47% in critically ill children and adults, thus impacting hundreds of thousands of patients annually. Often neuromorbidity evolves undetected until after clinical manifestations emerge and is irreversible. Neuromorbidity can strike acutely, e.g. seizures, stroke, intracerebral hemorrhage, cerebral edema, and/or delirium, or in a more protracted fashion, e.g. neuromuscular weakness and/or cognitive decline, and is often permanent, endured throughout the remainder of a person’s lifetime. No standard clinical tool exists to identify patients at risk for neuromorbidity or for real-time neurologic monitoring, in stark contrast to the heart, kidney, liver, and many other organs. To fill this gap and transform the way clinicians detect and monitor for evolving brain injury, we developed a Bio-digital Rapid Alert to Identify Neuromorbidity (BRAIN) that continuously feeds electronic health record (EHR) variables in 9 clinical domains (A through I) into proprietary informatics and machine learning platforms. Prototype BRAIN A-I models are robust and predict clinician concern for neuromorbidity before clinical action is taken. To link biological and digital signatures, we have defined a panel of serum biomarkers that can identify time-documented neuromorbidity before clinical detection. Using a “Bayesian to Bedside” approach, we have created a live data pipeline bridging the EHR and a dedicated host server, establishing the infrastructure necessary to operationalize BRAIN A-I as an embedded predictive analytic and decision-driving support tool. In this proposal we will test the hypothesis that digital signatures in the EHR coupled with brain-specific biomarkers can rapidly detect neuromorbidity in critically ill children. Successful deployment of interoperable, 24/7 point-of-care neurologic monitoring for early detection of neuromorbidity would represent a breakthrough for the clinical management of critically ill patients.