Mayo Clinic HeartShare Clinical Center

项目摘要/摘要 这是梅奥诊所申请以心脏共享的身份加入NIH心脏共享研究联盟 临床中心(CC)。我们的目标是与其他心脏共享调查人员合作，以澄清 射血分数保留的心力衰竭的病理生理学和新的诊断 和治疗方法。多种病理生理过程可能最终导致不同的HFpEF 表型，尽管具体的机制在很大程度上仍不清楚。也不知道是否标准 临床信息可以识别具有不同机制病因的患者，这是必要的，以提供 靶向治疗正在进行临床试验，并最终进入临床实践。我们的提案概述了四个具体目标。在……里面 具体目标1：我们记录了梅奥诊所拥有资源，梅奥心脏共享团队拥有 在HFpEF和相关疾病、临床研究、患者方面的专业知识和生产力记录 招聘和保留、数据科学和协作团队科学，帮助推动 红心共享网络。在具体目标2：我们提出了一个广泛的机械性表型鉴定协议，提供了 反映衰老、全身性疾病过程和多器官完整性的量化变量(L2 数据)，在无监督机器学习(ML)模型中用作输入变量。我们假设 这种方法将允许识别独特的HFpEF病理生理表型群(群)。我们也 提出有创性血流动力学特征，循环生物标记物和心肌的经心脏梯度， 脂肪和骨骼肌组织特征(L3数据)在每个HFpEF内的子集中获得 病理生理表型群。我们假设这些L3数据将增强靶向治疗的识别 战略。最后，我们概述了使用EHR数据来开发自动化算法以准确地实现有监督ML 确定使用L2数据得出的HeartShare HFpEF病理生理表型组。我们假设如果 如果成功，这种方法将通过允许自动识别来增强HeartShare发现的翻译 不同HFpEF表型组的患者参加针对其特异性药物的临床试验 病理生理学。在具体目标3中：我们建议单独使用循环蛋白(n=5000；由 基于SOMAScanTM适配子的平台)作为非监督ML模型的输入变量 HFpEF病理生理表型(簇)。具体目标4：我们概述了梅奥心脏共享研究 技能发展计划。为HFpEF临床研究人员提供短期密集沉浸 与数据科学家实习生在梅奥心血管疾病AI实习生或 作为梅奥·克恩中心数据科学学者的长期致力于数据科学的项目将配备 新一代HFpEF调查人员拥有强大的数据科学工具箱，以推动未来的发现。

PROJECT SUMMARY/ABSTRACT This is Mayo Clinic’s application to participate in the NIH HeartShare Research Consortium as a HeartShare Clinical Center (CC). Our goal is to collaborate with the other HeartShare Investigators to elucidate the pathophysiology of heart failure (HF) with preserved ejection fraction (HFpEF) and discover novel diagnostic and therapeutic approaches. Multiple pathophysiologic processes may ultimately lead to different HFpEF phenotypes, though the specific mechanisms remain largely undefined. It is also not known whether standard clinical information can identify patients with different mechanistic etiologies, which is necessary to provide targeted therapies in clinical trials and eventually in clinical practice. Our proposal outlines four specific aims. In Specific Aim 1: We document that Mayo Clinic has the resources and the Mayo HeartShare Team has the expertise and track record of productivity in HFpEF and relevant related diseases, clinical research, patient recruitment and retention, data science, and collaborative team science to help drive the success of HeartShare Network. In Specific Aim 2: We propose a broad mechanistic phenotyping protocol providing quantitative variables reflective of senescence, systemic disease processes, and multi-organ integrity (L2 data), which are used as input variables in unsupervised machine learning (ML) models. We hypothesize that this approach will allow identification of unique HFpEF pathophysiologic phenogroups (clusters). We also propose invasive hemodynamic signatures, trans-cardiac gradients of circulating biomarkers and myocardial, adipose and skeletal muscle tissue characterization (L3 data) be obtained in a subset within each HFpEF pathophysiologic phenogroup. We hypothesize these L3 data will enhance identification of targeted therapeutic strategies. Lastly, we outline supervised ML using EHR data to develop automatable algorithms to accurately identify the HeartShare HFpEF pathophysiological phenogroups derived using L2 data. We hypothesize that if successful, this approach will enhance translation of HeartShare findings by allowing automated identification of patients in the different HFpEF phenogroups for enrollment in clinical trials of agents targeting their specific pathophysiology. In Specific Aim 3: We propose that use of circulating proteins alone (n=5000; defined by the SOMAScanTM Aptamer based platform) as input variables for unsupervised ML models will identify unique HFpEF pathophysiologic phenotypes (clusters). In Specific Aim 4: We outline the Mayo HeartShare Research Skills Development Program. Providing HFpEF clinical investigators a short-term intensive immersion experience by collaboration with a data scientist intern in the Mayo Cardiovascular Disease AI Internship or a long term dedicated program in data science as a Mayo Kern Center Scholar in Data Science will equip a new generation of HFpEF investigators with a robust data science toolbox to drive future discovery.