Inteligencia Artificial en la Prevención de Accidentes Laborales: una Revisión Exhaustiva del Estado del Arte y Perspectivas Futuras

Abstract

To gain a deeper understanding of the role of artificial intelligence in preventing workplace accidents, a systematic search was conducted in the Scopus and ScienceDirect databases. In the first stage, approximately 900 articles related to the topic were identified. Subsequently, filters were applied by publication year (2024–2025), by author to avoid duplication, and by availability, selecting only studies whose full text could be opened and analyzed. This refinement yielded a current and accurate sample, from which 36 key studies were selected. These studies addressed practical applications, methodological approaches, and relevant results in various sectors, including transportation, industry, critical processes, occupational health, and discussions on ethics and trust in AI systems. The methodology consisted of classifying the articles according to their field of application and examining in detail the technological approaches employed. Among the most prominent techniques were machine learning, deep neural networks, multimodal systems, digital twins, and explainable artificial intelligence frameworks. The findings were organized into comparative tables that facilitated the identification of trends, technological advancements, and common challenges present in the various sectors analyzed. The results show that artificial intelligence has made significant progress on several fronts. Advances are evident in the early prediction of industrial failures, the strengthening of reliability in critical systems, the optimization of safety in work environments, and the integration of ethical and human elements into automated decision-making. These advances reflect a solid path toward safer, more efficient, and more responsible management of workplace risks.