Semi-Parametric Subgroup Analysis for Longitudinal Data with Applications to Multidisciplinary Approach to the Study of Chronic Pelvic Pain (MAPP) Study

项目摘要 本项目的目标是开发新的统计方法，将纵向/功能轨迹聚类为 亚组，并开发预测模型的集群成员使用基线和随时间变化的协变量。 所提出的方法的动机，并将被应用到，在NIDDK资助的数据收集 多学科方法研究慢性盆腔疼痛（MAPP）研究网络。这是一个持续 纵向队列研究，收集纵向泌尿系统慢性盆腔疼痛综合征（UCPPS）症状数据， 以及许多其他生物标志物、神经成像数据和微生物组数据。研究的目的是识别风险 可以预测特定患者未来UCPPS症状是否会恶化或改善的因素， 了解潜在的病理机制，并制定预防性治疗。我们将首先开发半- 纵向/功能性数据的参数分类和聚类方法，将同时考虑平均值 轨迹和随时间变化的变化。然后，我们将方法扩展到多元函数 设置，在其中我们将同时执行纵向因素分析，减少所有的纵向 将症状分解为更小维度的因素，并基于所有潜在因素对受试者进行聚类。第三 具体目标是制定随时间变化分类和聚类方法。我们还提出了一个在线监测 一种算法，该算法将现有的群体信息结合在检测新主题的切换中， 他的累积病史和随时间变化的风险因素。这可能会导致早期的医疗干预。所有的 建议的方法将附有用户友好的软件包，并将应用于收集的数据 正在进行的MAPP研究网络研究。

PROJECT SUMMARY The goal of this project is to develop novel statistical methods to cluster longitudinal/functional trajectories into subgroups, and to develop predictive models for cluster membership using both baseline and time-varying covariates. The proposed methods are motivated by, and will be applied to, the data collected in the NIDDK-funded Multidisciplinary Approach to the Study of Chronic Pelvic Pain (MAPP) Research Network. This is an ongoing longitudinal cohort study that collects longitudinal urological chronic pelvic pain syndrome (UCPPS) symptom data, together with many other biomarkers, neuroimaging data and microbiome data. The goal of the study is identify risk factors that can predict whether the future UCPPS symptoms for a specific patient will either worsen or improve, to understand the underlying pathological mechanisms and to develop preventive treatments. We will first develop semi- parametric classification and clustering methods for longitudinal/functional data that will take into account both mean trajectories and time-varying variabilities in the clustering. We will then extend the methods to multivariate functional settings, in which we will simultaneously perform longitudinal factor analysis that reduces all the longitudinal symptoms into smaller dimensional factors, and cluster the subjects based on all the underlying factors. The third specific aim will develop time-varying classification and clustering methods. We also propose an online monitoring algorithm that will incorporate the existing population information in detecting the switching of a new subject based on his cumulative history and time-varying risk factors. This hopefully could lead to early medical interventions. All the proposed methods will be accompanied with user-friendly software packages, and will be applied to the data collected from the ongoing MAPP Research Network Studies.