DATA MANAGEMENT AND STATISTICAL ANALYSIS CORE

数据管理和统计分析核心

• 批准号: 7493862
• 负责人: MARGARET RUTH BURCHINAL
• 金额: $ 9.49万
• 依托单位: UNIV OF NORTH CAROLINA CHAPEL HILL
• 依托单位国家: 美国
• 财政年份: 2008
• 资助国家: 美国
• 起止时间: 2008-07-01 至 2013-06-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7493862
• 关键词: Address; Algorithms; Analysis of Variance; Behavior assessment; Bioinformatics; Biometry; Books; Budgets; Characteristics; Charge; Child; Child Development; Cluster Analysis; Code; Cognitive; Collaborations; Complex; Computer software; Computers; Conceptions; Consult; Consultations; Core Facility; Data; Data Analyses; Data Collection; Data Set; Databases; Depth; Detection; Development; Discrimination; Documentation; Education; Ensure; Equation; Experimental Designs; Failure; Feedback; Funding; Genes; Genomics; Genotype; Goals; Grant; Growth; Human Resources; Image Analysis; Individual; Institutes; Licensing; Life Cycle Stages; Linear Models; Logistic Regressions; Measures; Mental Retardation; Methodology; Methods; Modeling; Numbers; Paper; Parents; Phase; Pilot Projects; Population; Procedures; Process; Productivity; Property; Psychometrics; Published Comment; Questionnaires; Range; Recommendation; Recruitment Activity; Regression Analysis; Research; Research Design; Research Personnel; Research Project Grants; Resources; Role; Sampling; Schools; Score; Screening procedure; Services; Site; Staging; Statistical Computing; Statistical Methods; Students; Supervision; Survival Analysis; System; Techniques; Technology; Testing; Time; Training; Travel; Valid Value; Work; Writing; base; computerized; concept; data integrity; data management; data mining; data modeling; design; direct application; experience; file format; genetic analysis; genetic association; instrument; interest; laptop; lectures; member; preference; quality assurance; repository; skills; social; teacher

A. OBJECTIVE Consultation on study design: The Data Management and Statistical Analysis (DMSA) Core provides consultation to investigators who are designing new studies. We prefer to become involved in the initial development of studies to assist in study design. We assist by providing feedback on the degree to which the proposed design can address the research questions with the proposed sample. We conduct power analyses to identify optimal study designs. We assist the investigator in developing the analysis plan by making recommendations regarding statistical methods to ensure that their analysis plan provides the most appropriate testing of research questions with the proposed instruments and samples. Finally, we can assist UNC DDRC investigators by providing feedback on instruments. We have over 25 years of experience in processing and analyzing assessments of children's social and cognitive skills and of parent characteristics and responsiveness with the child and can make recommendations regarding the psychometric properties of most widely used measures. Based on our experience, we can inform investigators whether the instruments they selected were successful (or unsuccessful) in previous studies with specific populations. For example, we found that it is extremely difficult to use some standardized cognitive screening instruments with children with moderate or severe mental retardation due to failure to obtain basal scores with most of those children and that some widely used teacher and parent questionnaires have questionable internal consistencies when used with individuals without posthigh school education. Consultation on data management: The application analyst provides consultation regarding data management to all UNC DDRC investigators. Careful implementation of quality assurance practices and documentation are necessary for maintaining the integrity of data as they are entered into databases. We meet with project staff to assist them in implementing recommended practices such as: ¿ Creating tracking logs to document the process of all data from collection through processing ¿ Creating data entry practices such as development of screens in which invalid or out-of-range values are tagged and data entry by two independent keyers and discrepancies are resolved by reviewing forms ¿ Creating carefully documented database files, including computerized error detection procedures (valid value/range, consistency), annotated code books and documents of decision rules for resolving errors and scoring data ¿ Consultation on instrument selection and the availability of computerized scoring algorithms for those instruments Consultation on statistical analysis: The statisticians in the Core provide statistical consultation to all UNC DDRC investigators. We are available to consult with all UNC DDRC regarding the analysis plans for funded UNC DDRC projects. This consultation can involve designing analysis plans and assisting UNC DDRC project staff in implementing those analyses and in interpreting results. Core statisticians have extensive experience with most methodologies used by UNC DDRC projects, including: Data reduction Regression, analysis of variance and linear models Logistic regression and survival analysis Growth curve analyses using linear and nonlinear methodologies Structural equation models Cluster analysis and discrimination methods useful in analyzing data from gene micro-array studies Analysis of differential expression or similar analysis on a variety of-omics platforms Analysis of genetic association data Model-based and spatial methods used to analyze imaging data Additional bioinformatics issues The statisticians and bioinformatics personnel assist UNC DDRC investigators by providing bioinformatics/ biostatistics support for basic experimental design and database management, guidance in the analysis of data, and data mining. Details of specific approaches are given further below. Core personnel provide assistance in interfacing with UNC genomics core microarray facilities, which primarily use Agilent and Affymetrix technologies. Genotyping core facilities have been greatly expanded in the last year, and are primarily based on the Illumina platform. DDRC researchers occasionally require assistance in interpreting basic file formats for their work. Core personnel can also aid in data sharing and uploading data to public repositories. A portion of internal data sharing of microarray data is mainly accomplished via the UNC Microarray Database, which has a log-in system and a common set of data formats. Training: The Core will continue providing seminars and short-courses to UNC DDRC investigators and project staff. We have offered one to two seminars per year on topics related to data management and statistical analysis as part of the UNC DDRC monthly investigators' meeting and plan to continue this. In addition, we planned to offer one short-course per year to project staff and interested investigators to provide an overview on data management practices and complex statistical methods, but found that working collaboratively with investigators using their project data was more effective. Unfortunately, investigators were too busy to participate in short-courses, whereas they found that we were quite effective in training when the training involved direct application of methods to their data. In these informal trainings, we consult with investigator regarding study design, data management, and data analysis. These consultations often introduce UNC DDRC investigators and staff to new practices and methodologies. We try to ensure that each investigator or staff member understands the basic concepts involved in and when these methods are appropriate for all practices we recommend. Similar training occurs for bioinformatics consultation and collaborations. Dr. Wright has given periodic lectures about statistical techniques for analysis of genetic data. UNC Center for Bioinformatics (http://bioinformatics.unc.edu) is an extremely valuable resource, and provides first-line basic training at no charge to UNC investigators, staff, and students. The Center for Bioinformatics Center has dedicated facilities, or can travel to an investigator's lab using laptop computers. Additionally the Center for Bioinformatics maintains numerous site licenses for bioinformatics analysis software (e.g., Genespring). Additional training of UNC investigators occurs in one-on-one consultation with Core statisticians. Hiring, training and supervision of project data managers and statisticians: We meet with interested UNC DDRC investigators to determine the extent to which they would like assistance with data management and statistical analysis. Consulting is available to all projects, but many investigators prefer to have the Core provide all data management and statistical support for their project. The Core is located within the Design and Statistical Computing Unit (DSCU) at the Frank Porter Graham Child Development Institute, and in the UNC Department of Biostatistics. The DSCU consists of 20 programmers and six statisticians. The Core is able to assign these programmers and statisticians to work on UNC DDRC projects if those projects pay for the time those individuals spend on their projects when requested services go beyond those supported by the Core. The easy availability of these programmers and statisticians for this expanded service provision is convenient for investigators and ensures that consistent high quality is maintained. We have been able to provide these services by using UNC DDRC funds for the recruiting, training, and supervision of DSCU staff who provide the actual data management and statistical analysis for the UNC DDRC projects. Investigators who request that our unit provide direct data management or statistical analysis beyond the consulting role described above that is considered part of the Core support must fund that effort from their grants. These DSCU staff members become part of the research team, and are hired, trained, and supervised by the Core personnel. The Core is able to recruit talented individuals and train them in professional data management and statistical methodologies. The Core supervises these individuals to ensure that quality assurance practices are maintained in data management and that appropriate and powerful statistical methods are used to analyze data. This is quite beneficial to projects with extensive behavioral assessments, especially since almost all projects are too small to need a full-time data manager or statistician and would be unable to recruit professionally trained individuals to work on their projects. Assistance with data management and analysis: Direct services are provided to assist investigators seeking additional funding. Core programmers and statisticians provide direct assistance to DDRC investigators as they develop new UNC DDRC projects or to help them maintain longitudinal projects when those projects are without other support. We assist investigators who conduct pilot studies by processing and analyzing those data. Core personnel also have provided data management and statistical analysis assistance to DDRC projects that are seeking renewal and may be without support at that time. This assistance has been especially valuable for longitudinal projects because data collection usually extends to the end of the funding cycle. With our assistance, the renewal proposal can include results from work funded by the last grant. Unfortunately, we cannot provide direct services to most projects due to the number of projects and the large amount of data collected by almost all projects. Our goal is to ensure that all investigators can be assisted when writing proposals, all projects can consult with us regarding data management or analysis, and any investigator can use our professional data managers and statisticians as part of their research team if the project pays for those direct services.

A.目标 研究设计咨询： 数据管理和统计分析（DMSA）核心向研究人员提供咨询 设计新研究。我们宁愿参与研究的初步发展，以协助研究 设计。我们通过提供有关拟议设计可以解决的程度的反馈来提供的协助 提出的样本研究问题。我们进行功率分析以识别最佳研究设计。 我们通过提出有关统计数据的建议来协助研究者制定分析计划 确保其分析计划提供最适当的研究问题测试的方法 建议的仪器和样品。最后，我们可以通过提供反馈来帮助UNC DDRC调查人员 乐器。我们在处理和分析儿童的评估方面拥有超过25年的经验 社会和认知能力以及父母的特征和对孩子的反应能力，可以使 有关大多数使用措施的心理测量特性的建议。基于我们 经验，我们可以通知调查人员他们选择的工具是否成功（或 在先前对特定人群的研究中，失败。例如，我们发现这非常困难 使用一些具有中度或重度的儿童使用一些标准化的认知筛查工具 由于未能与大多数孩子获得基本分数而导致的迟钝，并且一些广泛使用的老师 父母问卷与没有posthy的个人一起使用时具有可疑的内部一致性 学校教育。 关于数据管理的咨询： 应用分析师向所有UNC DDRC调查人员提供有关数据管理的咨询。 仔细实施质量保证实践和文档对于维护 数据进入数据库时​​的完整性。我们会见项目人员，以协助他们实施 推荐的做法，例如： »创建跟踪日志以记录从集合到处理的所有数据的过程 �创建数据输入实践，例如无效或范围值的屏幕开发 通过审查表格解决了两个独立的键音和差异的标记和数据输入 �创建仔细记录的数据库文件，包括计算机化错误检测过程（有效 价值/范围，一致性），针对解决错误的决策规则的注释代码书籍和文件 评分数据 „有关仪器选择的咨询和计算机评分算法的可用性 乐器 统计分析咨询： 核心中的统计数据为所有UNC DDRC调查人员提供了统计咨询。我们可以使用 就资助的UNC DDRC项目的分析计划咨询所有UNC DDRC。这项咨询可以 涉及设计分析计划并协助UNC DDRC项目人员实施这些分析和 解释结果。核心统计学家在UNC使用的大多数方法中都有丰富的经验 DDRC项目，包括： 减少数据 回归，方差分析和线性模型 逻辑回归和生存分析 使用线性和非线性方法进行生长曲线分析 结构方程模型 聚类分析和歧视方法可用于基因微阵列研究的分析数据 分析差异表达或在多种词平台上的类似分析 遗传关联数据分析 用于分析成像数据的基于模型和空间方法 其他生物信息学问题 统计和生物信息学人员通过提供生物信息学/ 生物统计学支持基本的实验设计和数据库管理，指导分析 数据和数据挖掘。特定方法的详细信息将在下面进一步提供。核心人员提供 协助与UNC基因组学核心微阵列设施接口，这些设施主要使用Agilent和 Affymetrix技术。去年，基因分型核心设施大大扩展了， DDRC研究人员偶尔需要帮助解释 他们的工作基本文件格式。核心人员还可以帮助数据共享并将数据上传到公众 存储库。微阵列数据的内部数据共享的一部分主要是通过UNC完成的 微阵列数据库，该数据库具有登录系统和一组通用的数据格式。 训练： 核心将继续向UNC DDRC调查人员和项目人员提供下水道和缩短服务。我们 每年都提供与数据管理和统计分析相关的主题的一到两个半手 UNC DDRC每月调查人员会议的一部分，并计划继续此次。此外，我们计划 每年为项目人员和有趣的调查人员提供一条短菜，以提供数据概述 管理实践和复杂的统计方法，但发现与 使用其项目数据的调查人员更有效。不幸的是，调查人员太忙了 参加短巡回赛，而他们发现我们在培训时在训练时非常有效 涉及将方法直接应用于其数据。在这些非正式培训中，我们咨询调查员 考虑研究设计，数据管理和数据分析。这些咨询经常引入UNC DDRC 调查人员和工作人员进行新的实践和方法。我们试图确保每个调查员或员工 成员了解涉及的基本概念以及这些方法适合所有实践 我们建议。 生物信息学咨询和合作也发生了类似的培训。赖特博士进行了定期讲座 关于用于分析遗传数据的统计技术。 UNC生物信息学中心 （http://bioinformatics.unc.edu）是一种非常宝贵的资源，并提供一线基本培训 向UNC调查人员，员工和学生收取费用。生物信息学中心中心有专门的设施， 或可以使用笔记本计算机前往调查员实验室。另外是生物信息学中心 维护许多用于生物信息学分析软件（例如Genespring）的站点许可证。额外的培训 UNC研究人员在与核心统计学家一对一咨询中发生。 项目数据经理和统计学家的雇用，培训和监督： 我们会见有趣的UNC DDRC调查人员，以确定他们想要的帮助程度 通过数据管理和统计分析。所有项目都可以咨询，但是许多调查人员 希望拥有核心为其项目提供所有数据管理和统计支持。核心是 位于Frank Porter Graham Child的设计和统计计算单元（DSCU）内 开发研究所以及UNC生物统计局。 DSCU由20名程序员组成 和六位统计学家。核心能够将这些程序员和统计学家分配为UNC DDRC 项目如果这些项目支付这些个人在要求服务时花在项目上的时间 超越了核心支持的那些。这些程序员和统计学家的易用性 扩大的服务提供对于调查人员来说很方便，并确保一致的高质量是 维护。我们已经能够通过使用UNC DDRC资金进行招聘来提供这些服务， 培训和对DSCU员工的监督，他们提供实际数据管理和统计分析 UNC DDRC项目。 要求我们单位提供直接数据管理或统计分析的调查人员 上面描述的咨询角色被认为是核心支持的一部分，必须为他们的努力提供资金 赠款。这些DSCU工作人员成为研究团队的一部分，并受到雇用，培训和监督 由核心人员。核心能够招募有才华的人并培训他们的专业数据 管理和统计方法。核心监督这些人，以确保质量 保证实践在数据管理中维护和适当有力的统计方法 用于分析数据。这对具有广泛行为评估的项目非常有益，尤其是 由于几乎所有项目都太小，无法需要专职数据经理或统计学家，并且将无法 招募经过专业培训的个人从事自己的项目。 协助数据管理和分析： 提供直接服务以帮助寻求额外资金的调查人员。核心程序员和 统计数据为DDRC调查人员开发新的UNC DDRC项目或向 当这些项目没有其他支持时，帮助他们维护纵向项目。我们协助调查人员 他们通过处理和分析这些数据进行试点研究。核心人员也提供了数据 管理和统计分析协助向寻求续订的DDRC项目，可能是 当时没有支持。对于纵向项目，这种帮助特别有价值，因为 数据收集通常扩展到资金周期的结束。在我们的协助下，续签建议可以 包括上次赠款资助的工作结果。 不幸的是，由于项目数量和大型项目，我们无法为大多数项目提供直接服务 几乎所有项目收集的数据量。我们的目标是确保所有调查人员都可以得到协助 在编写建议时，所有项目都可以就数据管理或分析以及任何项目咨询 研究人员可以使用我们的专业数据经理和统计学家作为研究团队的一部分 项目为这些直接服务支付。