Dynamic Longitudinal Functional Models with Applications to the CRIC Study

动态纵向函数模型及其在CRIC研究中的应用 该项目的目标是开发新的动态纵向函数模型并应用于 将其转换为随时间重复测量的心电图仪数据 慢性肾脏疾病(CKD)患者慢性肾功能不全队列(CRIC)研究。 我们还将开发实时风险预测算法，以识别高危人群 心血管疾病(CVD)。CRIC研究是对慢性阻塞性肺疾病患者的持续研究。 肾脏疾病(CKD)，由国家糖尿病、消化和肾脏研究所资助 疾病(NIDDK)自2001年以来。CRIC的研究记录了标准的12个领先 从七个临床中心招募的所有参与者每年进行一次心电图检查。我们的 主要目的是评估纵向心电模式是否是心血管疾病的先兆和 因此可以用来识别高危个体。我们提出了新的统计方法来提取 原始心电轨迹在基线和纵向变化方面的新特征 可以预测心血管并发症，如心力衰竭(HF)住院， 心肌梗死(MI)、中风、房颤(AFib)和心血管死亡。这个 信息将被纳入拟议的实时、计算高效的风险预测 算法。我们还将使用收集的CKD患者的外部队列来验证我们的发现 来自宾夕法尼亚大学医疗系统(UPHS)。 限于心电数据的分析，有着广泛的应用。我们将开发用户- 友好的软件包，用于新的统计模型和风险预测算法，并分享 验证数据，以促进其在统计和临床社区的使用。

Dynamic Longitudinal Functional Models with Applications to the CRIC Study The goals of this project are to develop novel dynamic longitudinal functional models and to apply them to the electrocardiographic (ECG) data that are measured repeatedly over time in the Chronic Renal Insufficiency Cohort (CRIC) study of individuals with chronic kidney disease (CKD). We will also develop real-time risk prediction algorithms to identify individuals at high-risk of cardiovascular diseases (CVD). The CRIC study is an ongoing study of individuals with chronic kidney diseases (CKD), funded by the National Institute of Diabetes, Digestive, and Kidney Diseases (NIDDK) since 2001. The CRIC study has recorded standard twelve-lead electrocardiograms (ECG) annually in all participants recruited from seven clinical centers. Our primary objective is to evaluate whether longitudinal ECG patterns are precursors to CVD and thus can be used to identity high-risk individuals. We propose new statistical methods to extract novel features from the raw ECG tracing both at baseline and in terms of longitudinal changes that are predictive of complications from CVD such as hospitalizations for heart failure (HF), myocardial infarction (MI), stroke, atrial fibrillation (AFib), and cardiovascular death. The information will be incorporated in the proposed real-time, computationally efficient risk prediction algorithms. We will also validate our discovery using an external cohort of CKD patients collected from the University of Pennsylvania Health System (UPHS).The proposed methods are not restricted to ECG data analysis and have a wide range of applications. We will develop user- friendly software packages for the new statistical models and risk prediction algorithms and share the validation data to promote their use in both statistical and clinical communities.