Intelligent Learner Digital Twin: From Multidimensional Monitoring to Adaptive Learning

Digital learning systems collect large amounts of data about learners, yet they often describe a student through a small set of indicators such as grades, attendance, and activity counts. These measures are useful, but they do not explain why a learner is struggling, losing focus, or becoming confused. This study proposes an intelligent learner digital twin that represents the learner through a dynamic and multidimensional state rather than a static profile. A focused review of learner modelling, multimodal learning analytics, adaptive learning, and digital-twin research was used to identify the main design requirements. Based on these requirements, a five-layer conceptual architecture is introduced. It covers learner-data collection, data processing and integration, AI-supported state inference, persistent digital-twin management, and adaptive feedback. The learner state includes cognitive, behavioural, attentional, and affective dimensions, together with a temporal history and confidence values. A five-level maturity model is also proposed to describe the transition from a static digital record to a predictive and closed-loop adaptive twin. The framework is examined through an online-learning scenario in which changes in attention and confusion can be detected before a clear decline in academic performance appears. The main contribution is a coherent link between educational data, AI-based state estimation, a persistent learner twin, and explainable educational action.

کلیدواژه ها:

learner digital twin ، artificial intelligence in education ، multimodal learning analytics ، adaptive learning ، learner modelling ، personalized education