Hybrid deep learning and IoT-based intelligent learning analytics system for personalized education

Personalized education has emerged as a transformative approach to enhancing student learning outcomes by adapting instructional content to individual needs and behaviors. The integration of IoT and hybrid deep learning enables continuous monitoring and intelligent analysis of learner activities in real time. However, existing learning analytics systems suffer from limitations such as centralized data processing, privacy concerns, limited adaptability, and an inability to effectively utilize multimodal data. These challenges reduce the accuracy of predictions and hinder dynamic personalization in diverse learning environments. To address these issues, this paper proposes a novel framework, Federated Multi-Modal Reinforcement Learning with IoT (FMMRL-IoT). The framework integrates IoT-based data collection with federated learning to ensure privacy-preserving distributed model training, while deep reinforcement learning enables adaptive decision-making for personalized learning paths. Multi-modal data fusion enhances the representation of student behavior and learning context. The proposed method continuously monitors student engagement, predicts learning performance, and dynamically adjusts content delivery, assessment strategies, and instructional pace in real time. It supports scalable deployment across smart classrooms and remote learning platforms. Experimental findings demonstrate that the proposed framework improves prediction accuracy, enhances student engagement, ensures data privacy, and significantly increases learning efficiency compared to traditional methods. Experimental results demonstrate that FMMRL-IoT framework can attain the prediction accuracy of 92–93%, user satisfaction of 88–92%, privacy preservation efficiency of 85–91%, adaption response time of 125 ms and computational complexity of ~ 2.3 s, outperform the existing personalized learning approaches.