Westborough High School Mental Health Predictive Analyses

Borghesani, Brian

Student Work

Westborough High School Mental Health Predictive Analyses

公开 Deposited

Background The rise in mental health issues among adolescents significantly threatens the well-being of many students. Early identification and intervention is crucial for developing intervention and treatment options. Many attributes have been studied to help identify factors that contribute to mental health problems but there is still no definitive model to flag students who are most at risk. Machine learning has been shown to be an effective way to analyze data from individuals and identify at-risk students who could benefit from personalized intervention. Methods Data was collected from 1009 students including 520 middle school (7th and 8th grade) and 489 high school students (9th and 11th grade) from the Westborough, MA public school district during the 2022-2023 academic year. Data included self-reported mental health survey data using the Revised Children’s Anxiety and Depression Scale (RCADS-25) for the middle school students and the Patient Health Questionnaire (PHQ-9) and the Generalized Anxiety Disorder scale (GAD-7) for the high school students. The data was analyzed using the Random Forest Classifiers (RFC) against demographic data (gender and race) and academic data (attendance, tardiness, early dismissals, grades, and whether they were on a English Language Learner (ELL), Special Education (SPED), Individual Education Program (IEP), and/or 504 plans). Data were analyzed to identify which demographic and academic factors predict whether students will score 10 or higher on the PHQ-9 or GAD-7, a recognized cutoff for clinical significance, indicating a binary risk classification (Kiely, 2015). This means students are categorized into two groups: those at risk (score 10 or higher) and those not at risk (score below 10). Results Using the risk binary classification, the model accurately predicted 85%-95% of students who scored high for both depression and anxiety. For middle school students both gender and race matter as both anxiety and depression scores (T-scores) showed clear differences across both gender and race. Additionally, for Asian students, the grade average showed importance for depression and dismissals showed importance for anxiety. For high school students, gender had a much stronger influence on predicting depression. However, for anxiety, gender and race along with grade average and average course level showed a clear difference in scores. Conclusion The results of this study, along with other research on WPS mental health data, show that Random Forest Classifiers (RFCs) hold promise for predictive analysis in identifying at-risk students. By identifying which demographic and academic factors are most associated with a GAD-7 or PHQ-9 score of 10 or higher, this data could be collected and used. Anxiety and depression T-scores varied across both gender and race. These findings align with prior research and underscore the importance of collecting demographic data along with self-reported data to identify students at risk for anxiety or depression.

This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator