<![CDATA[Elementary science instruction in the digital era often lacks interactivity and responsiveness to diverse learning needs, particularly for abstract topics such as the water cycle. This study examines the effect of adaptive animated media supported by deep learning features on Grade 4 students’ conceptual understanding and compares it with conventional instruction.A quasi-experimental nonrandomized control-group pretest–posttest design was employed with 50 fourth-grade students at SDN Sukarahayu 02 (25 experimental; 25 control). Both groups completed a 20-item multiple-choice pretest and posttest, while the experimental group also responded to a perception questionnaire. The experimental group used adaptive animated modules featuring progress monitoring, branching reinforcement, differentiated practice, and targeted feedback, whereas the control group received lecture-based instruction with textbooks and worksheets. Data were analyzed using normalized gain scores and the Mann–Whitney test.Initial abilities were comparable (experimental = 52.35; control = 51.80). Posttest scores were higher in the experimental group (82.75) than in the control group (68.45). Learning gains were greater in the experimental group (g = 0.638, medium) compared to the control group (g = 0.345, medium). The difference was statistically significant (p = 0.012). Student perceptions indicated increased engagement (88%), improved understanding (84%), enhanced visualization (88%), and higher motivation (80%).Adaptive animated media effectively supports conceptual understanding by providing personalized learning pathways and visualizing abstract processes, leading to more meaningful learning experiences than conventional methods.Deep learning–supported adaptive animation significantly enhances elementary students’ understanding of the water cycle and offers a more interactive and effective alternative to traditional instruction.]]>
