<![CDATA[As artificial intelligence (AI) increasingly reshapes digital education, Deep Learning (DL) has emerged as a pivotal technological paradigm to drive adaptive learning environments; however, a comprehensive synthesis tracing the algorithmic architectures that successfully foster these personalized systems remains sparse. This study aims to systematically investigate the foundational concepts, computational applications, and implementation challenges of DL models within modern educational frameworks. Adhering to the PRISMA protocol, a rigorous systematic literature review was conducted by screening prominent electronic databases, analyzing 127 peer-reviewed publications spanning from 2015 to 2024. The synthesis reveals that personalized learning constitutes the primary application focus of educational DL, accounting for 29.1% of the total literature. Computationally, advanced architectures such as Transformers and BERT demonstrate the highest operational success rate (91.4%) in facilitating context-aware intelligent tutoring and automated assessments. Conversely, the deployment of these technologies is significantly hindered by socio-technical constraints, notably technological infrastructure deficits (68.5%) and substantial gaps in educators' digital literacy (62.2%). Beyond mapping the current state-of-the-art, this research contributes an integrated socio-technical framework that bridges computational capability with pedagogical design, offering actionable, evidence-based guidelines tailored for educational stakeholders to systematically navigate infrastructure limitations and optimize AI-driven personalization.]]>
Copyrights © 2026
