Unmet educational accommodation needs and mental health outcomes in adults with disabilities: A machine learning approach.

Background: No research has yet determined exactly what accommodation needs are unmet for disabled students and how those needs being unmet predict psychosocial outcomes many years later.

Objective: To address this research gap, we seek to explore the potentially long-term associations of unmet educational accommodation needs and demographic characteristics with the mental health of adults with disabilities (n = 409).

Methods: To explore these associations, we use modern the machine learning technique of Random Forest feature importance.

Results: While 52.3 % of the sample reported having had one or more unmet accommodation needs while going to school, 57.2 % displayed current clinically elevated symptoms of depression and 48.4 % clinically elevated symptoms of anxiety. The machine learning approaches had 65.9 % and 60.0 % accuracy in correctly classifying clinically elevated depression and anxiety symptoms, respectively. For the models predicting clinically elevated depression symptoms using mean decrease in impurity (MDI) and permutation importance, unmet accommodation needs ranked fifth and fourth, respectively, in feature importance after age, disability severity, high school GPA, and individual income (for MDI). For the MDI model predicting clinically elevated anxiety symptoms, unmet academic accommodation ranked third in feature importance behind disability severity and age, while for permutation importance, unmet academic accommodation need ranked fourth behind age, urbanicity, and disability severity.

Conclusion: Unmet academic accommodations may result in reduced psychological adjustment and quality of life potentially many years into adulthood.