An Integrated Pedagogical–Technical Framework for Deep Learning in Vocational Education and Training: A Systematic Review

Abstract* Background Vocational Education and Training (VET) institutions play a crucial role in welcoming the era of Industry and Society 5.0, where the integration of cyber-physical technology requires the mastery of employability skills that include adaptability, work readiness, and transversal competencies. Deep Learning, as an advanced data-based computational learning system, offers transformative potential for redesigning vocational learning processes. Methods This study employed a Systematic Literature Review (SLR) method following the PRISMA protocol. A total of 16 articles were identified, screened, and analyzed in depth to synthesize the pedagogical and technical parameters of Deep Learning in VET and to develop an integrated conceptual framework. Results Key findings identify that the most effective Deep Learning parameters are not only technical (learning rate, epoch), but depend on their integration with pedagogical frameworks. The key parameters that emerge are the ability of Deep Learning models to support: (1) mastery learning through adaptive learning pathways; (2) contextual learning by presenting real-world industry problems; (3) authentic learning through smart work practice simulations; and (4) immersive learning in responsive virtual environments. Conclusions The implications of this research confirm that the application of Deep Learning in VET institutions is no longer sufficient to simply adopt the technology, but must be pedagogically integrated through a model structure specifically designed to strengthen authentic and competency-based learning. This model is positioned to enable graduates to independently update their occupational skills through a Deep Learning-based adaptive system.