MCA: Leveraging Artificial Intelligence to Enhance the Creativity of the STEM Professional Workforce by Transforming Education for Neurodiverse Learners

该项目由职业中期促进计划资助，该计划支持科学家和工程师通过协同和互惠互利的合作伙伴关系实质性地加强他们的研究计划。这个MCA项目采用了一种变革性的方法，通过增加STEM学科中神经多样性学生的参与来满足国家对创造性专业劳动力的迫切需求。要应对我国面临的大规模、复杂、多方面的挑战，科技突破至关重要。神经科学生无与伦比的天赋和独特的能力为我们国家解决这些困难提供了一个独特的机会。然而，我们的STEM教育对文本内容的过度依赖和STEM文本的语言复杂性可能会脱离并阻碍神经多样性的个人，如阅读困难的人。该项目利用了从多个以前由NSF资助的项目中获得的独特知识和专业知识，以及一支强大的多学科研究团队，他们在神经认知科学、神经成像、STEM教育、阅读障碍和人工智能(AI)方面拥有专业知识。这项大胆、趋同的研究建立在人工智能的最新进展以及神经成像技术提供的机会的基础上，为患有诵读困难的中学生提供了一种可扩展、可个性化的文本简化工具。该项目将走在向个性化辅助工具转变的前沿，以加强不同神经的学生在STEM教育中的参与。该项目有可能显著提高阅读障碍中学生的学习能力，尽管他们在空间可视化和发散思维方面具有独特的天赋，但他们在STEM领域的代表性仍然很低。虽然这个项目的重点是阅读困难的个人，但拟议的框架可能会进一步扩大到包括其他群体，如自闭症学生和英语学习者(ELS)。该项目包括三个整合的活动，以促进知识和生成关键数据，为阅读困难学生开发个性化的辅助工具。该项目的目标是：1)采用大型预先训练的自然语言处理(NLP)模型来完成STEM文本简化任务；2)开发来自阅读困难学生的大型反馈数据集，并使用强化学习(RL)来定制该模型；以及3)使用脑电(EEG)数据的特征来测量与阅读理解相关的神经认知功能，从而使模型个性化。在阅读任务中收集的学生反馈数据集和个体脑电可能会对未来阅读障碍、神经认知科学、STEM教育和机器学习的研究有很大的帮助。这项研究的结果可能会鼓励使用大脑数据进行个性化学习的进步。通过这个项目形成的研究伙伴关系促进了旨在提高不同神经学学生的学术成功的大规模多学科研究的发展。一名行业协作者将参与该项目的早期工作，以确保广泛传播辅助工具和项目成果。该项目预计将通过消除参与的障碍，增进学业和社会弱势群体的福祉。该项目的方法可能会显著推动个性化辅助学习技术的发展，并为独特的学生在我们的传统教育系统中取得成功提供新的机会。该奖项反映了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。

This project is funded by the Mid-Career Advancement program that supports opportunities for scientists and engineers to substantively enhance their research program through synergistic and mutually beneficial partnerships. This MCA project embraces a transformative approach to address the critical national need for a creative professional workforce by increasing the participation of neurodiverse students in STEM disciplines. Scientific and technological breakthroughs are crucial to address the large-scale, complex, and multifaceted challenges facing our nation. The unparalleled talents and unique abilities of neurodiverse students present a distinctive opportunity for our nation to address these difficulties. However, the excessive reliance of our STEM education on textual content and the linguistic complexities of STEM texts may disengage and discourage neurodiverse individuals, such as those with dyslexia. This project capitalizes on the unique knowledge and expertise gained from multiple previously NSF-funded projects and a strong multidisciplinary research team with expertise in neurocognitive science, neuroimaging, STEM education, dyslexia, and artificial intelligence (AI). This bold, convergent research builds on the latest advancements in AI as well as opportunities provided by neuroimaging technologies to advance a scalable, personalizable text simplification tool for middle school students with dyslexia. This project will be at the forefront of the shift towards personalized assistive tools to enhance the participation of neurodiverse students in STEM education. The project has the potential to significantly enhance the learning of middle school students with dyslexia, who despite their unique talents in spatial visualization and divergent thinking, remain highly underrepresented in STEM fields. Though this project is focused on individuals with dyslexia, the proposed framework may be further broadened to include other groups, like students with Autism, and English learners (ELs).This project includes three integrated activities to advance knowledge and generate critical data for the development of a personalized assistive tool for students with dyslexia. This project aims to: 1) adapt large pretrained natural language processing (NLP) models for STEM text simplification tasks, 2) develop a large dataset of feedback from students with dyslexia and use reinforcement learning (RL) to customize the model for this population, and 3) use features of electroencephalogram (EEG) data to measure neurocognitive functions related to reading comprehension to personalize the model. The student feedback dataset and individual EEGs collected during the reading tasks may significantly benefit future research in dyslexia, neurocognitive sciences, STEM education, and machine learning. The outcomes of the study may encourage the advancement of personalized learning using brain data. The research partnership formed through this project facilitates the development of large-scale multidisciplinary studies aimed at enhancing the academic success of neurodiverse students. An industry collaborator will be involved early in the project to ensure a broad dissemination of the assistive tool and project outcomes. This project is expected to enhance the well-being of an academically and socially vulnerable group of students by dismantling barriers to participation. The approach of this project may significantly advance the development of personalized assistive learning technologies and present new opportunities for unique students to succeed in our traditional education system.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.