AI-Based Learning Analytics and Real-Time Monitoring for IoT-Enabled Smart Laboratory of Engineering Education

Abstract

Engineering laboratories play a significant role in facilitating problem-based learning across electrical, environmental, computer, and interdisciplinary engineering programs; however, most lab sessions still rely on physical attendance, delayed grading, limited equipment references, and little visibility into student engagement. This lack of visibility compromises instructors’ ability to identify learning barriers, unsafe conditions, uneven resource use, and at-risk students during lab sessions. An Internet of Things (IoT) smart lab framework is proposed that uses real-time sensor data, edge processing, and cloud storage to support learning analytics for improved operational and pedagogical decisions. The framework integrates data from environmental sensors, smart equipment, workstation logs, access control, and learning platforms to create a continuous data stream. An analytic layer processes this data using structured event learning and sequence modeling to estimate engagement, detect at-risk sessions, identify equipment bottlenecks, and enable timely intervention. A prototype simulated undergraduate engineering lab is developed to evaluate latency, reliability, predictive performance, and interpretability. Results show 93.4% engagement accuracy, 0.914 macro-F1, 0.887 AUROC for early risk detection, and 1.8 seconds median latency, outperforming temporal and non-temporal baselines. These findings demonstrate that multimodal sensing with explainable AI improves operational visibility and educational analytics. The framework defines the engineering lab as a cyber-physical learning environment rather than a collection of devices, supporting safer, more efficient, and data-driven learning across engineering disciplines.