项目名称: 新型统计模型在精神疾病的基因、脑影像和行为数据整合中的应用
项目编号: No.11471081
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 数理科学和化学
项目作者: 罗强
作者单位: 复旦大学
项目金额: 71万元
中文摘要: 本项目旨在提出新的统计模型来分析精神疾病研究中的基因、脑影像、行为这三个尺度的数据,主要研究以下三个方面的问题:(1)大脑各个脑区之间相互关联,简单把脑影像看成独立的随机变量来处理,人为造成信息损失,我们希望提出一种统计模型描述不同条件下各个脑区间的时空依赖关系,并从时空依赖关系的角度来对比分析正常人和病人;(2)已有证据表明大脑的某些活动受遗传信息的影响,但是全基因组和全脑影像之间的关联分析遇到了高维复杂数据组带来的诸多挑战,我们希望充分利用数据的时空结构对该问题进行降维,并提高关联分析的统计检定力,发现新的致病基因;(3)目前通过分析影像-基因和影像-行为之间的联系来推断行为-基因联系的主流研究方法存在明显局限,我们希望建立严格的统计模型同时整合基因-影像-行为数据,定量地描述这三个尺度之间的关系,寻找横跨三个尺度的生物标记,提高精神疾病的诊断准确率。
中文关键词: 数据处理;网络分析;影像遗传学;关联分析;贝叶斯模型
英文摘要: This project aims at developing new statistical models to integrate genetic, neuroimaging, and behavioural data for mental disorders. We will tackle the following problems: (1) Considering the neuroimaging data as a group of independent variables loses information due to the interaction between different parts of the brain. We will build a statistical model to describe the spatio-temporal dependency among brain regions, and try to discriminate patients from healthy controls in this aspect;(2)Converging evidences suggest that various of brain activities are inheritable, but the whole-genome and whole-brain association study encounters many challenges as both data are in high-dimensions and with complex structures. We will develop new statistical models for the dimension-reduction in this association study, in order to improve its statistical power and to discover new pathogenic genes; (3)Clear pitfalls lie in the current approach of associating gene with behaviour through the two separate associations, including gene-neuroimaging and neuroimaging-behaviour. We will integrate the data in those three scales by a rigorous statistical model at the same time, expecting this method will help us to build new bio-markers across those three scales for mental disorders.
英文关键词: Data analysis;network analysis;imaging genetics;association analysis;Bayesian model