<![CDATA[Science education requires a pedagogical shift from knowledge transmission toward competency development that prepares students for complex twenty-first-century challenges. This study systematically reviewed the implementation of deep learning approaches in science education and their contribution to students’ twenty-first-century skills. The review followed the PRISMA framework through identification, screening, eligibility assessment, and inclusion. Searches conducted in Scopus and Google Scholar identified 505 records published between 2016 and 2025, of which 19 articles met the inclusion criteria. Data were extracted according to publication year, country, educational level, instructional strategy, targeted competencies, and principal findings. Results showed a clear increase in publications, with Indonesia contributing the largest number of studies and higher education representing the dominant research setting. Various strategies, including project-based learning, problem-based learning, inquiry-based learning, STEAM, blended learning, flipped classrooms, and self-directed learning, were used to operationalize deep learning principles. These strategies consistently supported critical thinking, creativity, communication, collaboration, self-regulation, self-efficacy, and student engagement. The findings confirm that deep learning functions as an integrated pedagogical orientation that connects conceptual understanding, authentic application, active participation, and reflection. Greater research attention is needed at primary and secondary levels, alongside longitudinal and cross-cultural studies examining sustained competency development. The review provides practical guidance for curriculum design and teacher development.]]>
Copyrights © 2026
