Understanding COVID-19 infection among people with intellectual and developmental disabilities using machine learning

Area of Science:

• Public Health
• Machine Learning
• Health Disparities

Background:

• People with intellectual and developmental disabilities (IDD) faced disproportionate COVID-19 impacts.
• Predicting COVID-19 infection in the IDD population presents significant challenges.

Purpose of the Study:

• To evaluate a machine learning model for predicting COVID-19 diagnosis in individuals with IDD.
• To identify key predictors of COVID-19 diagnosis among Home and Community Based Services (HCBS) users with IDD.

Main Methods:

• Merged three major IDD-specific datasets, including over 700 variables.
• Developed a random forest machine learning algorithm to predict COVID-19 diagnosis.
• Analyzed a random sample of HCBS users within a single state.

Main Results:

• The machine learning model achieved 62.5% accuracy in predicting COVID-19 diagnosis.
• Primary predictors included higher age, extensive support needs (overall, medical, behavioral), lower-income neighborhoods, higher Medicaid expenditure, and elevated BMI.
• Findings align with general population trends, indicating increased social contact may heighten COVID-19 risk.

Conclusions:

• Machine learning offers utility in predicting COVID-19 in IDD populations.
• Demographic and support-related factors significantly influence COVID-19 risk.
• Understanding these predictors can inform targeted public health interventions for vulnerable groups.