Integrative modeling and dynamic prediction of Alzheimer's disease

项目总结/摘要 拟议的R 01赠款是对PAR-18-352“在全球范围内的方法和测量”的直接回应。 行为和社会科学（R 01）。阿尔茨海默病（Alzheimer's disease，AD）是一种进行性神经退行性疾病 导致多个域中的损伤（例如，认知、行为和生活质量）和进展 在时间上、跨领域和跨个人上是不均匀的。没有单一的生物标志物提供足够的 信息，以捕捉整个疾病谱的潜在严重程度。因此，AD研究收集 来自多个源的数据（例如，临床、神经成像和遗传;多模态数据）。我们提出了一种新 综合建模框架，提供了原则性推理，准确的个性化预测， 疾病进展，以及基于新的受试者特异性数据的动态预测更新。这种新颖的模式 发展是重要的，以确定风险和保护因素的AD和目标的高风险个人，以及 以便个性化管理、预后和治疗选择。总体目标是：（1） 开发一个多变量功能混合模型（MFMM）的纵向临床综合建模 数据;（2）使用这种模型来提供对目标的未来结果轨迹和风险的个性化预测 事件;（3）通过整合高维神经影像学和遗传学， 数据;（4）通过专业软件开发和网络使这种方法易于访问 部署.我们的方法可以广泛应用于其他具有类似多模态数据的临床研究 结构

Project Summary/Abstract The proposed R01 grant is in direct response to PAR-18-352 “Methodology and Measurement in the Behavioral and Social Sciences (R01)”. Alzheimer's disease (AD) is a progressive, neurodegenerative disorder that causes impairment in multiple domains (e.g., cognition, behavior, and quality of life) and progresses heterogeneously in time and across domains and individuals. No single biomarker provides sufficient information to capture the underlying severity of disease across the entire spectrum. Hence, AD studies collect data from multiple sources (e.g., clinical, neuroimaging, and genetic; multi-modal data). We propose a novel integrative modeling framework to provide statistically-principled inference, accurate personalized prediction of disease progression, and dynamic prediction update, based on new subject-specific data. This novel model development is important to identify risk and protective factors for AD and target high risk individuals, as well as to personalize the management, prognosis, and treatment selections. The overall objectives are to: (1) develop a multivariate functional mixed model (MFMM) for the integrative modeling of the longitudinal clinical data; (2) use such model to provide personalized prediction of future outcome trajectories and risks of target events; (3) advance the integrative model by incorporating the high-dimensional neuroimaging and genetic data; (4) make this methodology easily accessible via professional software development and web deployment. Our methods can be broadly applied to other clinical studies with similar multi-modal data structure.