Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{187461,
        author = {S. Vimala and Dr.G. Arockia Sahaya Sheela},
        title = {Real-Time Smartphone Distraction Detection in Virtual Learning via Attention-CNN-LSTM},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {6},
        pages = {5644-5656},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=187461},
        abstract = {Objective: This study addresses the challenge of smartphone-driven distractions during synchronous online learning by developing an intelligent system that detects distraction episodes in real time while prioritizing learner privacy and maintaining computational efficiency.  Approach: We engineered an attention-enhanced CNN-LSTM architecture that combines Convolutional Neural Networks for spatial feature extraction with Long Short-Term Memory layers for temporal pattern recognition. The addition of a self-attention mechanism enables the model to highlight which behavioral signals most strongly indicate distraction, supporting explainable decision-making without requiring invasive surveillance technologies.  Results: Testing with 120 university students across approximately 180 hours of virtual learning sessions yielded encouraging outcomes: 92.4% accuracy, 0.91 F1-score, and a processing latency of just 1.2 seconds per analysis window. The attention-enhanced model substantially outperformed baseline approaches, including standard CNN-LSTM, LSTM-only, and classical machine learning algorithms. Visualization of attention weights confirmed that the model focuses on meaningful behavioral indicators—such as rapid touch sequences and accelerometer spikes—rather than arbitrary statistical patterns.  Innovation: This research contributes a practical, privacy-respecting deep learning framework that connects device-level activity monitoring with educational analytics. The system executes on-device, eliminating the need to transmit sensitive data to external servers, and integrates interpretability directly into its architecture. These qualities position it as a viable tool for real-time deployment in digital classrooms.},
        keywords = {Smartphone Distraction, Virtual Learning, CNN-LSTM Attention, Real-time Detection, Explainable AI, Privacy-Preserving Machine Learning, Educational Data Mining},
        month = {November},
        }