KNOWLEDGE MANAGEMENT FOR CLINICAL TRIAL ADVICE SYSTEMS

我们提出了医学计算研究，以探索知识的方法 综合咨询系统的采购和知识库管理， 特别是在随机临床试验领域。ONCOCIN，An 正在进行的由美国国立卫生研究院支持的研究项目，以开发一个建议系统 肿瘤学的临床试验，将提供初步开发 这些知识管理工具的环境。我们的目标是发展 两种主要类型的交互工具： 1.允许医生设计新的研究方案的工具 并将预定义的协议直接输入计算机，而不需要 需要依赖计算机程序员(知识工程师)来 援助。生成的知识库随后将由ONCOCIN使用 当它为患者提供建议和数据管理支持时 这些新的协议。 2.允许知识工程师维护知识库的工具， 探索其内容，对无法录入的专业知识进行编码 并创建供专家使用的新程序 希望在医疗领域以外的领域进入实验方案 肿瘤学。 在开发供内科专家使用的互动工具时，我们将 开发三个子方案：(1.1)知识获取系统(称为 Opal)允许肿瘤学家为ONCOCIN输入新的方案知识 在没有计算机科学家帮助的情况下，(1.2)用于绘制图形的工具 审查现有协议信息，以及(1.3)自动化软件 生成描述编码协议的文本文档。这些 计划将允许ONCOCIN和ONCOCIN的新肿瘤学方案快速进入 也将作为肿瘤学家开发的“计算机辅助设计”工具 新协议。 在开发供知识工程师使用的互动工具时，我们将 还开发了三个子程序：(2.1)基于图形的编辑器以供审查 和修改协议知识，以便甚至可以输入概念 如果不能使用知识获取系统对它们进行编码，(2.2) 允许扩展知识获取系统的例程 医生可能会输入以前无法指定的概念，以及(2.3)系统 创建全新的知识获取系统用于其他类型的 临床试验。

We propose medical computing research to explore methods of knowledge acquisition and knowledge base management for integrated advice systems, specifically in the area of randomized clinical trials. ONCOCIN, an ongoing NIH-supported research project to develop an advice system for clinical trials in oncology, will provide the initial development environment for these knowledge management tools. Our goal is to develop interactive tools of two major types: 1. Tools that will allow physicians both to design new research protocols and to enter predefined protocols directly into the computer, without needing to rely on a computer programmer (knowledge engineer) for assistance. The resulting knowledge base will then be used by the ONCOCIN program when it provides advice and data management support for patients on these new protocols. 2. Tools that will allow a knowledge engineer to maintain a knowledge base, explore its contents, encode specialized knowledge not able to be entered directly by the physician, and create new programs for use by experts wishing to enter experimental protocols in medical domains other than oncology. In developing the interactive tools for use by physician experts, we will develop three subprograms: (1.1) a knowledge acquisition system (known as OPAL) to allow oncologists to enter new protocol knowledge for ONCOCIN without assistance from computer scientists, (1.2) a facility for graphical review of existing protocol information, and (1.3) software for automatic generation of text documents that describe encoded protocols. These programs will permit rapid entry of new oncology protocols for ONCOCIN and will also serve as "computer aided design" tools for oncologists developing new protocols. In developing the interactive tools for use by knowledge engineers, we will also develop three subprograms: (2.1) a graphics-based editor for review and modification of protocol knowledge so that concepts may be entered even if they cannot be encoded using the knowledge acquisition system, (2.2) routines to permit extension of the knowledge acquisition system so that physicians may enter previously unspecifiable concepts, and (2.3) a system for creating entirely new knowledge acquisition systems for other types of clinical trials.