Statistical Methods for Ultrahigh-dimensional Biomedical Data

超高维生物医学数据的统计方法

基本信息

批准号：
9900790
负责人：
Jianqing Fan
金额：
$ 29.3万
依托单位：
PRINCETON UNIVERSITY
依托单位国家：
美国
项目类别：
财政年份：
2006
资助国家：
美国
起止时间：
2006-02-01 至 2022-01-31
项目状态：
已结题

来源：
https://reporter.nih.gov/project-details/9900790
关键词：
Address Alzheimer&apos s Disease Big Data Big Data Methods Biological Biomedical Research Brain Classification Clinical Communication Computer software Cox Models Cox Proportional Hazards Models Data Data Set Databases Dependence Dimensions Disease Disease Progression Evaluation Gene Expression Gene Proteins Genes Genomics Heterogeneity Internet Investigation Learning Linear Models Location Meta-Analysis Methods Molecular Outcome Ownership Patients Polynomial Models Principal Component Analysis Privacy Proteins Proteomics Research Role Statistical Data Interpretation Statistical Methods Tail Techniques Testing Time autism spectrum disorder big biomedical data bioinformatics tool cell type computing resources feature selection genetic information health data high dimensionality high throughput analysis improved machine learning method macrophage model building next generation novel precision medicine predict clinical outcome simulation statistical and machine learning statistics therapeutic target tool transcriptome sequencing treatment effect

项目摘要

This proposal develops novel statistics and machine learning methods for distributed analysis of big data in biomedical studies and precision medicine and for selecting a small group of molecules that are associated with biological and clinical outcomes from high-throughput data such as microarray, proteomic, and next generation sequence from biomedical research, especially for autism studies and Alzheimer’s disease research. It focuses on developing efficient distributed statistical methods for Big Data computing, storage, and communication, and for solving distributed health data collected at different locations that are hard to aggregate in meta-analysis due to privacy and ownership concerns. It develops both computationally and statistically efficient methods and valid statistical tools for exploring heterogeneity of big data in precision medicine, for studying associations of genomics and genetic information with clinical and biological outcomes, and for feature selection and model building in presence of errors-in- variables, endogeneity, and heavy-tail error distributions, and for predicting clinical outcomes and understanding molecular mechanisms. It introduces more robust and powerful statistical tests for selection of significant genes, SNPs, and proteins in presence of dependence of data, valid control of false discovery rate for dependent test statistics, and evaluation of treatment effects on a group of molecules. The strength and weakness of each proposed method will be critically analyzed via theoretical investigations and simulation studies. Related software will be developed for free dissemination. Data sets from ongoing autism research, Alzheimer’s disease, and other biomedical studies will be analyzed by using the newly developed methods and the results will be further biologically confirmed and investigated. The research findings will have strong impact on statistical analysis of high throughput big data for biomedical research and on understanding heterogeneity for precision medicine and molecular mechanisms of autism, Alzheimer’s disease, and other diseases.

该建议开发了新颖的统计数据和机器学习方法，用于分布式分析生物医学研究和精确医学的大数据，并选择一小部分与高通量数据的生物学和临床结果相关的分子例如微阵列，蛋白质组学和生物医学研究的下一代序列，特别是自闭症研究和阿尔茨海默氏病研究。它专注于发展大数据计算，存储和通信的有效分布式统计方法，并解决在不同位置收集的分布式健康数据，这些数据很难汇总由于隐私和所有权问题而引起的荟萃分析。它在计算和统计上有效的方法和有效的统计工具，用于探索大数据的异质性精确医学，用于研究基因组学和遗传信息与临床的关联以及生物学结果，以及在存在错误的情况下进行特征选择和模型构建变量，内生性和重尾误差分布，并用于预测临床结果并了解分子机制。它引入了更强大，更有力的统计数据在存在数据依赖性的情况下选择重要基因，SNP和蛋白质的测试，对依赖测试统计数据的错误发现率的有效控制和治疗评估对一组分子的影响。每种建议的方法的优势和弱点将是通过理论研究和仿真研究进行了严格的分析。相关软件将是自由传播开发。来自正在进行的自闭症研究，阿尔茨海默氏病的数据集，以及其他生物医学研究将通过使用新开发的方法和结果将在生物学上进一步确认和研究。研究结果将有对生物医学研究的高吞吐量大数据的统计分析的强烈影响了解精确医学的异质性和自闭症的分子机制，阿尔茨海默氏病和其他疾病。