The purpose of this study was to better understand the possible educational applications of artificial intelligence techniques. We employed a three-step data labeling process that used previously identified mental health symptoms (i.e., prior classification system), natural language processing, and reliability and validity feedback from content experts. Using unstructured student text data, we identified mental health symptoms of high school students with intellectual or developmental disabilities who had been accepted into to a postsecondary education transition program at a large Midwest university.

We analyzed Intellectual Assessments (IAs) and Individual Education Programs (IEPs) for 78 adults aged between 16 and 21 years old with an Intellectual or Developmental Disability from 2015 to 2022. Analysis of the records extracted 2062 quotations associated with 27 different mental health symptoms. Preliminary analysis revealed quotations were associated with attention issues (20%), anxiety (13%), depressive mood (6%), focus issues (6%), hyperactivity (6%), restlessness (6%), weight and appetite issues (5.9%), feeling lost (5%), lack of insight (4%), and sleeping issues (3.8%). Subjectivity analysis revealed that the text from documents submitted between 2020 and 2022 were more neutral and objective compared to the text found within the 2015 to 2019 documents. More text data were able to be pulled for male students compared to female students for both the IAs and IEPs. Students in this study showed that they have co-existing executive functioning needs and higher rates of anxiety and depression.

Project Team

Erica Kaldenberg, Associate Research Scientist - Realizing Education and Career Hopes (REACH)

Mingying Zheng, Graduate Assistant, PhD Candidate in Educational Measurement and Statistics