Improving Health Outcomes through Computer Generation of Reader-Appropriate Texts

许多研究表明，提供给消费者的健康信息的可读性并不 与他们的一般阅读水平相匹配(Rudd，Moeykens等人2000)。即使在医疗保健部门的努力下 提供者和作者使材料更具可读性，今天大多数以患者为导向的网站、小册子、 药品标签和出院说明仍要求消费者具有十级阅读水平或 Higher(Nielsen-Bohlman，Panzer等人)2004年)。然而，消费者的阅读水平并不少见 低到四年级。 为了解决这个问题，我们建议开发一种基于计算机的方法来提供适当的文本 消费者的可读性。统计自然语言处理技术的新进展(Barzilay 2003年；Barzilay和Elhada 2003；Elhada，McKeown等人。2005)使我们相信计算机程序 可以开发为显著增加可读内容的范围和数量 消费者。它还可能有助于改善理解、自我管理和最终的临床结果。 这个项目的总体目标是通过自动化的 从难懂到易懂的内容翻译。具体目标是 1.开发一种用于评估卫生文本可读性的计算机化工具。我们将加强 现有的可读性仪器(Zakaluk和Samuels，1988年3月1日)，包括测量 健康术语难度、文本衔接、内容组织和布局。 2.开发一款用于将复杂的健康文本翻译成新版本的工具，目标是 可读性水平，没有关键信息丢失，使用统计自然语言处理 技巧。 3.进行评估研究，以验证提供具有适当可读性的内容是否具有积极意义 对读者理解的影响。我们将使用普通内科药物作为我们系统的试验台 临床及其糖尿病患者。我们将为这些患者提供自我护理材料。 与公共卫生的相关性：拟定可读性评估工具和“平面图” 英语“工具将有助于将复杂的健康材料翻译成适合读者的文本。它具有以下潜力 使公众更容易获得卫生领域的大量可用信息。

Many studies have shown that the readability of the health information provided to consumers does not match their general reading levels (Rudd, Moeykens et al. 2000). Even with the efforts of healthcare providers and writers to make materials more readable, today most patient-oriented Web sites, pamphlets, drug-labels, and discharge instructions still require the consumer to have a tenth grade reading level or higher (Nielsen-Bohlman, Panzer et al. 2004). Not infrequently, however, consumer reading levels are as low as fourth grade. To address this problem, we propose developing a computer-based method for providing texts of appropriate readability to a consumer. Recent progress in statistical natural language processing techniques (Barzilay 2003; Barzilay and Elhadad 2003; Elhadad, McKeown et al. 2005) lead us to believe that computer programs can be developed to dramatically increase the range and amount of readable content available to consumers. It may also help improve comprehension, self management and eventually clinical outcome. The general goal of this project is to provide consumers with readable content through the automated translation of content from difficult to understand to easy to understand. The specific aims are 1. To develop a computerized instrument for assessing the readability of health texts. We will enhance existing readability instruments (Zakaluk and Samuels March 1, 1988) by including measurements of health term difficulty, text cohesion, and content organization and layout. 2. To develop a "Plain English" tool for translating complex health texts into new versions at targeted readability levels with no critical information loss, using statistical natural language processing techniques. 3. To conduct an evaluation study to verify that providing content with appropriate readability has a positive impact on reader comprehension. We will use as a test bed for our system a general internal medicine clinic and its diabetes patients. We will provide self-care materials to these patients. Relevance to public health: The proposed development of a readability assessment instrument and "Plain English" tool will help translating complex health materials into reader-appropriate texts. It has the potential of making the vast amount of available information in the health domain more accessible to the lay public.