| Learning analytics has gained significant attention as a means of extracting actionable insights from large volumes of educational data generated through learning management systems, mobile applications, and online platforms. However, the widespread adoption of learning analytics and AI-driven educational systems raises critical concerns related to data privacy, security, and ethical use. This paper proposes a security-aware and privacy-preserving learning analytics framework designed to support human-centered precision education by embedding security and privacy controls across the full educational data lifecycle. Additionally, the paper positions differential privacy as a core element of the privacy-preserving layer and analyzes how it can be deployed within learning analytics as an architectural space, distinguishing architectural settings (central, local, and federated), analytical tasks (descriptive, predictive, and data-sharing), and points of application (input, processing, and output level). The paper also highlights key challenges and design considerations for adopting differential privacy in practice. Successfully leveraging differential privacy in learning analytics requires communication among learners, instructors, administrators, and researchers to comprehend and agree on the privacy–utility trade-offs in order to sustain trust while preserving educational value. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.