<![CDATA[Background: The rapid expansion of artificial intelligence in education has transformed elementary mathematics instruction. However, most studies remain technology-centered and insufficiently examine how adaptive deep learning reshapes students’ conceptual understanding and pedagogical practice. Aims: This study synthesizes existing research to reframe adaptive deep learning as a pedagogical framework that supports conceptual understanding through data-informed personalized learning in elementary mathematics. Methods: A qualitative systematic literature review was conducted on peer-reviewed studies focusing on adaptive deep learning applications in mathematics education. Selected articles were analyzed using thematic synthesis to identify recurring pedagogical mechanisms, conceptual outcomes, and instructional implications. Results: The synthesis indicates that adaptive deep learning enhances conceptual understanding by diagnosing misconceptions, personalizing instructional pathways, and integrating dynamic visual representations of abstract concepts. Learning analytics generated by adaptive systems support teacher decision-making and formative assessment. The reviewed studies consistently report improvements in engagement, conceptual clarity, and learning equity, particularly in heterogeneous elementary classrooms. Conclusion: Adaptive deep learning should be conceptualized not merely as instructional technology but as a pedagogically grounded framework that restructures how conceptual understanding develops. When integrated with teacher mediation and curricular alignment, it fosters inclusive, responsive, and concept-oriented learning environments. Its transformative potential lies in harmonizing algorithmic adaptation with professional judgment, thereby strengthening both instructional precision and conceptual depth in elementary mathematics education.]]>
Copyrights © 2025
