TREE-BASED SOFTWARE FOR BIOMEDICAL APPLICATIONS

用于生物医学应用的基于树的软件

• 批准号: 6071793
• 负责人: JAMES SCHIMERT
• 金额: $ 37.5万
• 依托单位: INSIGHTFUL CORPORATION
• 依托单位国家: 美国
• 财政年份: 1998
• 资助国家: 美国
• 起止时间: 1998-07-01 至 2002-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6071793
• 关键词: computer data analysis; computer human interaction; computer program /software; computer system design /evaluation; data management; statistics /biometry

We propose to develop software for tree-based methods which implements research on biomedical applications not currently supported by commercial software. Examples include survival and longitudinal data analysis, clustering, and generalized tree models. Tree-based tools offer many advantages including easy interpretation. Tree representations may mimic the way that many medical scientists think about data. Trees are also relatively fast to build and search, making them suitable for interactive clustering of large data sets such as encountered in image analysis. Phase II research will develop algorithms, model selection strategies, and stability diagnostics for tree based methods. Dynamic, interactive graphical tools will enable the user to explore the resulting tree-based model, and interpret structure in data by linking the tree representation to a variety of graphical and analytical tools. We will create software that analysts find flexible and easy to use, enabling medical researchers to use tree-based tools to explore and understand data from a wide variety of applications. Additionally, the software will be supported in an integrated environment for data analysis, and permit analysts, consultants, and statistical researchers to extend the software to incorporate future innovations in recursive partitioning research. PROPOSED COMMERCIAL APPLICATIONS: This SBIR will result in a new software package which is inter-operable with existing S-Plus products. We expect this product to appeal to a wide market of users including data analysts, consultants, and researchers in disciplines as diverse as medicine, pattern recognition, data mining, and image analysis.

我们建议开发用于基于树木的方法的软件，以实现商业软件当前未支持的生物医学应用的研究。 例子包括生存和纵向数据分析，聚类和广义树模型。 基于树的工具提供了许多优势，包括简单的解释。 树的表征可能会模仿许多医学科学家对数据的看法。 树木也相对较快地构建和搜索，使其适用于图像分析中遇到的大型数据集的交互式聚类。第二阶段研究将开发基于树方法的算法，模型选择策略和稳定性诊断。 动态，交互式图形工具将使用户能够通过将树表示形式链接到各种图形和分析工具来探索所得的基于树的模型，并解释数据中的结构。我们将创建分析师发现灵活且易于使用的软件，使医学研究人员能够使用基于树的工具来探索和了解各种应用程序的数据。 此外，该软件将在集成环境中支持数据分析，并允许分析师，顾问和统计研究人员扩展该软件，以将未来的创新纳入递归分区研究中。拟议的商业应用程序：此SBIR将产生一个新的软件包，该软件包与现有S-Plus产品相互作用。 我们希望该产品能够吸引广泛的用户市场，包括数据分析师，顾问和研究人员，例如医学，模式识别，数据挖掘和图像分析。