Contribution of artificial intelligence research to the development of inclusive applications for people with disabilities: A scoping review (2015-2025).

Abstract

Artificial intelligence (AI) is rapidly transforming healthcare. Yet people with disabilities (PWD), representing 16% of the global population, remain disproportionately excluded from these advances. This exclusion is driven by algorithmic bias, non-representative datasets, and limited involvement of PWD in the design, development, and evaluation of AI applications. This scoping review examines the extent to which AI research in health contributes to the development of inclusive technologies for PWD, focusing on equity considerations. It identifies current trends, methodological gaps, and opportunities for socially responsive innovation. We conducted a systematic search across MEDLINE, Embase, Web of Science, and CINAHL for the 2015-2025 period, following Arksey and O'Malley's five-stage scoping review methodology and the PRISMA-ScR guidelines. Data were extracted and synthesized thematically using the Population-Concept-Context approach. Of 4,233 references retrieved, 93 studies were included. Most studies originated from high-income countries, particularly in North America (45.2%) and Europe (26.9%). AI applications addressed motor, sensory, cognitive, communication-related, and psychosocial disabilities, with emphasis on prediction/diagnosis, assistive technologies, and personalized care. While about 60% of studies referenced equity, only 12 (12.9%) implemented concrete measures (e.g., adapting tools for vulnerable groups or ensuring sample diversity), and only six (6.5%) actively involved PWD through co-design or user validation. AI holds significant potential to enhance autonomy, access, and quality of life for PWD. Realizing this promise will require moving beyond technical optimization to embrace inclusive design, participatory methods, and equity-driven frameworks aligned with responsible AI development and use.