Digital-Based Deep Learning in Elementary Science Education: Implementation and Outcomes in Ecosystem Learning

Deep learning, in the pedagogical sense—defined here as an instructional approach that engages students in exploration, discussion, and reflection to achieve meaningful conceptual understanding beyond surface memorization—has strong potential to transform elementary science education. This study describes the implementation of digital-based deep learning (D-DL) in Grade V science learning on the ecosystem topic at a public elementary school in Bekasi Regency, Indonesia. Using a qualitative descriptive approach, data were collected through observations, interviews (with the principal, teachers, and students), and documentation. Supporting quantitative data (pre-test/post-test scores and participation percentages) were analyzed descriptively (means and percentages). Digital media—simulation/experiment videos (concept exploration), interactive applications such as Wordwall (inquiry and formative feedback), and Google Classroom (collaborative reflection)—were used to facilitate the deep learning cycle. Participation rose from 35% to 81% and the mean test score from 63.2 to 83.7, with 86.4% of students achieving scores ≥70. Supporting factors included principal support and teacher readiness; constraining factors included limited devices, unequal digital skills, and connectivity issues. This study offers practical insights for implementing D-DL in elementary science aligned with Indonesia's Independent Curriculum