Transforming Resuscitation through Artificial INtelligence (TRAIN Study)

通过人工智能改变复苏（TRAIN 研究）

• 批准号: 10712407
• 负责人: THOMAS D REA
• 金额: $ 71.34万
• 依托单位: UNIVERSITY OF WASHINGTON
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-07-01 至 2028-05-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10712407
• 关键词: Accounting; Acute; Adult; Affect; Algorithms; Anti-Arrhythmia Agents; Artificial Intelligence; Blinded; Carbon Dioxide; Caring; Cessation of life; Chest; Childhood; Clinical; Decision Making; Defibrillators; Dose; Electric Countershock; Electrocardiogram; Experimental Models; Functional disorder; Goals; Heart Arrest; Heterogeneity; Hospitals; Hypoxia; Impedance Cardiography; Individual; Intelligence; Interruption; Investigation; Measures; Monitor; Morphologic artifacts; Morphology; Myocardial Ischemia; Outcome; Patients; Performance; Perfusion; Phenotype; Physiologic pulse; Physiological; Physiology; Population; Process; Prognosis; Protocols documentation; Provider; Public Health; Research; Resuscitation; Signal Transduction; System; Techniques; Technology; Testing; Time; Translating; United States; Vasoconstrictor Agents; Ventricular Fibrillation; Vital Status; artificial intelligence method; blind; clinical application; clinical phenotype; computerized data processing; data resource; digital; electric impedance; improved; individual patient; individualized medicine; innovation; mortality; multidisciplinary; novel strategies; optimal treatments; out-of-hospital cardiac arrest; personalized approach; precision medicine; prevent; prognostic; prospective; randomized trial; signal processing

Project Summary Mortality from out-of-hospital sudden cardiac arrest (OHCA) is a large public health burden, accounting for approximately 10% of all deaths in the US. Because OHCA is a leading cause of mortality, advances in resuscitation have the potential to improve public health. Currently resuscitation protocols use a one-size-fits- all approach. However, we now understand that OHCA occurs via heterogeneous mechanisms and manifests a time-dependent pathophysiology, which influences prognosis. The heterogeneity suggests discrete clinical phenotypes and an opportunity for individualized therapy. Distinguishing information about patient physiology can be harnessed from the defibrillator. Continuous bio-measures of ECG, end-tidal carbon dioxide (ETCO2), and transthoracic impedance (TI) can determine physiologic status and potentially guide optimal treatment. However, a real-time continuous approach to characterize a patient’s physiology and prognosis by accurately determining the underlying rhythm and its vitality is not presently feasible without repeatedly interrupting CPR. CPR interruption is required because chest compressions introduce ECG artifact that prevents rhythm identification, prognostic assessment of rhythm morphology, and a patient’s underlying vital status (vitality). However, CPR interruption is harmful since it disrupts perfusion in the otherwise pulseless OHCA victim. Consequently, the current protocol is a compromise: CPR is interrupted every 2 minutes to help inform care decisions though treatment proceeds empirically as CPR resumes and providers are typically “blinded” to the actual underlying rhythm and vital status. Emerging evidence from the proposal team highlight the ability to use signal processing techniques to investigate the ECG, ETCO2, and TI defibrillator signals during CPR to improve OHCA resuscitation. These investigations use artificial intelligence (AI) methods to determine a patient’s instantaneous physiological status and predict downstream resuscitation outcomes. We propose an investigative plan that will: 1. Derive and validate an integrated ventricular fibrillation (VF) OHCA algorithm that incorporates and builds upon previously-validated modular algorithms, using artificial intelligence (AI) methods that process and integrate ECG, TI, and ETCO2 bio-signals during active CPR. 2. Prospectively evaluate the integrated algorithms and their validated building block components in distinct adult and pediatric OHCA populations. 3. Conduct a simulated randomized trial among EMS to compare the described precision strategy to the current-day, fixed protocol to understand how dynamic prompts of a precision strategy affect CPR metrics. The project leverages an unparalleled data resource and a tested, multidisciplinary team with a track-record of impactful resuscitation investigations involving novel approaches to AI and resuscitation. This consequent precision strategy ultimately could transform resuscitation and improve public health.

项目概要 院外心脏骤停 (OHCA) 造成的死亡是一项巨大的公共卫生负担， 约占美国所有死亡人数的 10%。由于 OHCA 是死亡的主要原因，因此在 复苏有可能改善公众健康。目前的复苏方案使用一刀切的方法 所有方法。然而，我们现在了解到 OHCA 通过异构机制和表现形式发生 时间依赖性病理生理学，影响预后。异质性表明离散的临床 表型和个体化治疗的机会。有关患者生理学的区分信息 可以从除颤器中利用。连续测量心电图、呼气末二氧化碳 (ETCO2)、 经胸阻抗（TI）可以确定生理状态并有可能指导最佳治疗。 然而，通过实时连续的方法来表征患者的生理学和预后 准确地确定潜在的节奏及其活力目前在不反复进行的情况下是不可行的 中断心肺复苏。需要中断心肺复苏，因为胸外按压会引入心电图伪影， 妨碍心律识别、心律形态的预后评估以及患者潜在的生命体征 状态（活力）。然而，心肺复苏中断是有害的，因为它会扰乱无脉搏的灌注。 OHCA 受害者。因此，当前的方案是一种折衷方案：每 2 分钟中断一次心肺复苏以帮助 尽管随着心肺复苏的恢复和提供者通常会根据经验进行治疗，但仍可告知护理决策 对实际的潜在节律和生命状态“视而不见”。 提案团队的新证据强调了使用信号处理技术来 在 CPR 期间研究 ECG、ETCO2 和 TI 除颤器信号，以改善 OHCA 复苏。这些 研究使用人工智能（AI）方法来确定患者的瞬时生理状态 并预测下游复苏结果。我们提出一项调查计划，该计划将： 1. 推导并验证集成心室颤动 (VF) OHCA 算法，该算法合并并构建了 基于先前验证的模块化算法，使用人工智能 (AI) 方法来处理和 在主动心肺复苏期间整合心电图、TI 和 ETCO2 生物信号。 2. 前瞻性地评估集成算法及其经过验证的构建块组件 成人和儿童 OHCA 人群。 3. 在 EMS 中进行模拟随机试验，将所描述的精确策略与 当前的固定协议，用于了解精确策略的动态提示如何影响 CPR 指标。 该项目利用了无与伦比的数据资源和经过测试、具有良好记录的多学科团队 涉及人工智能和复苏新方法的有效复苏调查。由此产生的 精准策略最终可以改变复苏并改善公共卫生。