Modelos de aprendizaje automático para la detección temprana de factores de riesgo en el desempeño académico de estudiantes universitarios mediante variables socio-tecnológicas
Tesla Revista Científica · 21.10.2026 · 6 (2) · e754
DOI: 10.55204/trc.v6i2.e754 ↗Kurz gefasst
Fachartikel in der Zeitschrift Tesla Revista Científica, online erschienen am 21.10.2026. Verfasst von José Fabián Cevallos Mazón und 5 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This study examines machine learning models applied to the early detection of risk factors in university students' academic performance, with particular emphasis on integrating socio-technological variables derived from Learning Management System (LMS) platforms. A systematic literature review was conducted following the PRISMA 2020 protocol, analyzing 40 studies published between 2020 and 2026 retrieved from international and Latin American databases. The main findings indicate that ensemble methods, namely Random Forest, XGBoost, and Gradient Boosting, represent the current state of the art in academic risk prediction, achieving accuracy values between 85% and 93% with an area under the curve (AUC-ROC) exceeding 0.90. Furthermore, the incorporation of digital behavior variables — LMS access frequency, session duration, forum participation, and assignment submission punctuality was found to increase the predictive power of models by 7 to 14 percentage points compared to the exclusive use of historical grades. A hybrid architecture integrating supervised predictive models with reinforcement learning agents is proposed for generating personalized pedagogical interventions in real time within institutional infrastructure.
Zitieren
APA
Mazón, J. F. C., Guashpa, W. F. T., Gavilanes, T. P. S., Yunda, W. A. C., Vaca, C. S. R., & Estrella, T. L. A. (2026). Modelos de aprendizaje automático para la detección temprana de factores de riesgo en el desempeño académico de estudiantes universitarios mediante variables socio-tecnológicas. Tesla Revista Científica, 6(2), e754. https://doi.org/10.55204/trc.v6i2.e754
Vancouver
Mazón JFC, Guashpa WFT, Gavilanes TPS, Yunda WAC, Vaca CSR, Estrella TLA. Modelos de aprendizaje automático para la detección temprana de factores de riesgo en el desempeño académico de estudiantes universitarios mediante variables socio-tecnológicas. Tesla Revista Científica. 2026;6(2):e754. doi:10.55204/trc.v6i2.e754
BibTeX
@article{Mazn2026,
author = {Mazón, José Fabián Cevallos and Guashpa, Walter Fabian Taday and Gavilanes, Tatiana Paola Sánchez and Yunda, Wilter Alexander Camacho and Vaca, Carlos Santiago Ramos and Estrella, Teresita Lourdes Argüello},
title = {Modelos de aprendizaje automático para la detección temprana de factores de riesgo en el desempeño académico de estudiantes universitarios mediante variables socio-tecnológicas},
journal = {Tesla Revista Científica},
year = {2026},
volume = {6},
number = {2},
pages = {e754},
doi = {10.55204/trc.v6i2.e754}
}
EUVIMED