A Socioeconomic Vulnerability Index for Persons with Disabilities via Machine Learning Approach

Employment is a key social determinant of health, yet people with disabilities (PWDs) face persistent disparities in vocational rehabilitation (VR) services and outcomes. Federally funded Vocational Rehabilitation (VR) programs serve individuals whose disabilities create barriers to competitive integrated employment, yet service allocation and outcomes vary widely across socioeconomic contexts. Understanding how cumulative socioeconomic disadvantage shapes VR service delivery and employment outcomes is essential for advancing equity. Using Rehabilitation Service Administration (RSA-9/11) data (n = 182,266), machine learning methods, including principal component analysis (PCA), were applied to develop a Socioeconomic Vulnerability Index, which was then used to examine associations with VR service delivery and employment outcomes. Results revealed that higher vulnerability was linked to increased likelihood of receiving basic support services but lower odds of career-oriented services. Higher vulnerability index scores correlated with higher expenditures (+ $246) yet substantially lower wages (- $3.13/hour). Findings suggest that an intake‑level vulnerability screening tool could support risk stratification and more targeted, equity‑focused VR service planning.