SBIR Phase I: A mobile, device-based, screening tool for assessing K-6 students’ cognitive and motor skills via machine learning handwriting analysis

这个小企业创新研究（SBIR）第一阶段项目的广泛影响是改善K-6学生的认知和运动技能评估。更具体地说，该项目将开发并执行一种新的、客观的、机器学习驱动的方法的可行性测试，以分析学生的书写熟练程度，这是一种衡量认知和运动技能的方法。通过使用智能设备应用程序，最终用户（教师、助手和家长）将能够拍摄学生的笔迹，并立即收到有关熟练程度、笔迹错误类型和有针对性的干预建议的结果。考虑到生成笔迹样本需要大量的视觉运动、精细运动和高阶认知技能，以及高达30%的学生有困难的事实，我们需要新的、更好的检测方案。随着COVID-19危机导致虚拟学习环境的使用增加，识别认知和运动技能缺陷变得尤为重要。学生们更多地使用电脑，老师们也减少了手写作业。这个小企业创新研究（SBIR）第一阶段项目的重点是开发机器学习（ML）算法，以生成高度准确、快速和客观的手写熟练程度预测。这些算法试图预测书写错误子类型。手写图像的ML分析以前从未做过。通过使用数据注释方案，在获取单个手写句子的单个照片后，智能设备应用程序将创建并访问高度敏感和特定年级的算法。这项技术被设想为一种通用的筛选工具，在每学年开始时用于识别书写水平欠佳的学生。手写熟练程度的实时分析将允许更早的识别和更早的干预，以提高学生的成绩，并为学区节省成本。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值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 of this Small Business Innovation Research (SBIR) Phase 1 project is to improve the assessment of cognitive and motor skills in K-6 students. More specifically, this project will develop and perform feasibility testing of a novel, objective, and machine learning-driven approach to analyzing student handwriting proficiency, a measure of cognitive and motor skills. Through the use of a smart device application, end-users (teachers, aides, and parents) will be able to take a photo of a student's handwriting and receive immediate results regarding proficiency, handwriting error types, and targeted intervention suggestions. Given the myriad of visual motor, fine motor, and higher-order cognitive skills needed to generate a handwriting sample and the fact that up to 30% of students have difficulties, there is a need for new and better detection schemes. The identification of cognitive and motor skill deficiencies is becoming especially important with the increased use of virtual learning environments due to the COVID-19 crisis. Students are interfacing more with computers, and teachers have decreased access to handwriting assignments.This Small Business Innovation Research (SBIR) Phase 1 project focuses on developing machine learning (ML) algorithms to generate highly accurate, rapid, and objective predictions of handwriting proficiency. These algorithms seek to predict the handwriting error sub-type. ML analysis of handwriting images has never been done before. Through the use of data annotation schemes, highly sensitive and grade-specific algorithms will be created and accessed by a smart device application following the acquisition of a single photo of a single handwritten sentence. This technology is envisioned as a universal screening tool to be used at the beginning of each school year to identify students with subpar handwriting proficiency. The real-time analysis of handwriting proficiency will allow for earlier identification and earlier interventions to improve student outcomes and deliver cost savings to school districts.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.