Development and Evaluation of a Virtual Aphasia Clinici*

描述（由申请人提供）：该项目的目标是开发和测试使用虚拟临床医生对失语症患者进行计算机自动化治疗的方法。我们建议将中风后语法障碍治疗项目的研究成果结合起来，该项目有相当多的可用疗效数据，即，潜在形式治疗（TUF） （Thompson等人，2003年），与一个发达且经过测试的交互式计算机系统相结合，该系统可以与科罗拉多大学口语研究中心（CSLR）开发的智能动画代理进行面对面交流。虽然这项工作的组成部分已经到位，但它们从未组合在一起，因此代表了创新研究。我们建议开发一种“虚拟失语症临床医生”，能够为失语症治疗提供刺激。该自动化系统将被评估为失语症患者独立使用的潜力，以进一步提高他们在语言恢复方面的进展，而不是在当前医疗政策下有限的临床接触所带来的预期结果。如果成功，这个项目将为下一步奠定基础——开发语音识别软件，允许虚拟临床医生评估失语症患者的语言反应。在项目的第一阶段，我们将使用目前可用的TUF协议来开发系统，以训练语法复杂的句子。第二阶段将实施和实验评估第一阶段开发的自动tuf -导师（TUF-T）协议的有效性。我们建议比较传统的，临床提供的TUF和两种TUF- t条件，一种有和一种没有临床医生调解。工作假设是，当考虑到时间和临床医生成本限制以及语言改善时，TUF- t将比临床医生管理的TUF更有效或同样有效。如果成功，这项工作将为结合和利用计算机技术的进步来管理言语和语言障碍带来新的方向。

DESCRIPTION (provided by applicant): The goal of this project is to develop and test a computer-automated treatment for individuals with aphasia using a virtual clinician. We propose to combine the research findings resulting from a program of work focused on treatment of post-stroke agrammatism, for which there are considerable available efficacy data, i.e., Treatment of Underlying Forms (TUF) (Thompson et al. 2003), with a well-developed and tested interactive computer system that enables face-to-face communication with an intelligent animated agent developed at the Center for Spoken Language Research (CSLR) at the University of Colorado. Although the components for this work are in place, they have never been combined and thus represent innovative research. We propose to develop a "virtual aphasia clinician" that will be capable of delivering stimuli for aphasia treatment. The automated system will be evaluated for its potential independent use by aphasic individuals to further their progress in language recovery beyond what might be expected to result from the limited clinical contact available under current healthcare policy. If successful, this project will lay the groundwork for a next step - to develop speech recognition software permitting the virtual clinician to evaluate aphasic users' verbal responses. In Phase I of the project we will develop the system using presently available TUF protocols for training syntactically complex sentences. Phase 2 will implement and experimentally assess the effectiveness of the automated TUF-tutor (TUF-T) protocols developed in Phase 1. We propose to compare traditional, clinician-delivered TUF to two TUF-T conditions, one with and one without clinician mediation. The working hypothesis is that TUF-T will be more effective than, or equally effective as, clinician-administered TUF when time and clinician-cost restraints, as well as language improvement are considered. If successful, this work will result in new directions for the management of speech and language disorders that incorporate and take advantage of advances in computer technology.