SBIR Phase I: Building a Novel, Interactive Artificial Intelligence Software Platform to Improve Communication in Autism

小型企业创新研究(SBIR)第一阶段项目的更广泛影响/商业潜力将是为言语障碍儿童，特别是自闭症谱系障碍儿童的照顾者和教育工作者创造一个工具。如果不加以治疗，这些障碍会对生活质量产生持久的影响，但治疗仍然困难且昂贵；据估计，全球自闭症治疗市场价值32亿美元。该项目将扩大在非临床环境中提供治疗的能力。将要开发的技术将包括在没有言语病理学家的情况下对言语发育水平进行自动评估。此外，还将开创一项技术，使活动和课程能够针对每个学习者进行单独定制和持续更新。这两项进步将构成一种新产品的技术核心，该产品专门为满足自闭症儿童的照顾者、教师和家庭的需求而设计。这个小型企业创新研究(SBIR)第一阶段项目涉及研究和开发治疗与自闭症谱系障碍相关的言语障碍的先进技术。人工智能将被应用于两个基本挑战，即在没有言语病理学家的情况下自动确定言语发育水平；以及根据用户的输入分配活动和课程。该项目将从语言学习者的不同发展阶段收集标签文本，以形成训练语料库。然后，机器学习将被用于构建评分引擎的算法，以使学习评估独立于临床医生。同时，业务规则管理系统的开发和应用将允许平台接受用户输入，如图像、单词或短语；它将输出新的单词和短语供学习者尝试。系统将遵循既考虑用户输入又考虑先前分配的语言开发级别的决策逻辑。这项研究的目标是产生与语言发展和决策逻辑的标签评估紧密匹配的算法，为每个级别分配适当的活动。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact /commercial potential of this Small Business Innovation Research (SBIR) Phase I project will be the creation of a tool for caregivers and educators of children with speech impairments, particularly those with autism spectrum disorder. Left untreated, these impairments have lasting impacts on quality of life but remain difficult and expensive to treat; the market for autism treatment is estimated at $3.2 B worldwide. This project will expand the capabilities to provide treatment in non-clinical settings. The technology to be developed will include automated evaluation of speech development levels in the absence of a speech pathologist. In addition, technology will be pioneered that allows activities and lessons to be individually customized and continuously updated for each learner. These two advances will form the technological core of a new product designed specifically to meet the needs of caretakers, teachers and families of children with autism. This Small Business Innovation Research (SBIR) Phase I project involves research and development of advanced technologies for the treatment of speech impairment associated with autism spectrum disorder. Artificial intelligence will be applied to two fundamental challenges, the automated determination of speech development levels in the absence of a speech pathologist; and the assignment of activities and lessons in response to user input. This project will collect labeled text from language learners at various stages of development to form a training corpus. Machine learning will then be used to build algorithms for scoring engines to make learning assessments independent of a clinician. In parallel, the development and application of a business rules management system will allow the platform to accept user input such as an image, word, or phrase; and it will output new words and phrases for the learner to attempt. The system will follow a decision logic that considers the both the user input and the previously assigned language development level. The goal of this research is to produce algorithms that closely match the labeled assessments of language development and decision logic that assigns activities appropriate for each level.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.