Systematic Review of University Academic Performance Supported by AI-Based Screen Readers

Correa A.J.C.; Espinoza L.Z.T.; Córdova E.H.A. · Proceedings of the 2025 IEEE 32nd International Conference on Electronics, Electrical Engineering and Computing, INTERCON 2025 (2025) · DOI: 10.1109/intercon67304.2025.11244635

Resumen

The implementation of screen readers based on artificial intelligence represented a promising alternative to reduce educational barriers faced by university students with visual impairments. These systems interpreted and presented complex digital content through speech synthesis and automated recognition. However, their integration into educational platforms remained limited, and the involvement of real users in their evaluation was scarce, which affected their effectiveness in various academic contexts. This paper analyzed the impact of these technologies on the academic performance of students with visual impairments, comparing them with traditional methods. To this end, a systematic review was conducted following the PICOC and PRISMA methodologies, identifying and analyzing 30 studies published between 2020 and 2025 in Scopus and ScienceDirect. The bibliometric analysis, supported by tools such as VOSviewer, enabled the identification of patterns, technologies used, evaluation metrics, and key contributions. The results evidenced improvements in the comprehension of complex content, digital navigation, and autonomy, highlighting tools such as Seeing AI and Envision, which demonstrated high effectiveness in text reading and speed. A growing acceptance of personalized virtual assistants and deep learning models with accuracy above 85% was also observed. It was concluded that these technologies contributed significantly to inclusion and the enhancement of academic performance, although challenges related to user participation, ethics, and the standardization of evaluations persisted.

Instituciones: Universidad Tecnológica del Perú