A Multimodal IoT Framework for Analyzing Student stress through Personalized Yoga Intervention and Explainable AI (XAI)

Abstract

Psychological stress among students is a critical public health issue, it affects academic performance and long-term mental stability. Traditional diagnostic methods like interviews and periodic counselling, which failed to provide continuous and personalized monitoring. This proposal introduces an automated framework Internet of Things (IoT) wearables and Artificial Intelligence (AI) to monitor physiological stress- to detect student stress in real time. The framework employs machine learning and deep learning models for multimodal stress prediction. It recommends personalized yoga interventions based on individual stress profiles. To address the "black box" nature of current deep learning architectures, we incorporate Explainable AI (XAI) techniques like SHAP and LIME. These used to ensure clinical transparency and user trust. This enables students and healthcare professionals to understand the factors influencing stress. The system is expected to enhance early stress detection, improve effectiveness, support decision making by healthcare professionals, and also encourage student engagement in preventive wellness practice. Performance may be evaluated using Accuracy, Precision, Recall, F1-score, Mean Absolute Error and Explainability Score. In addition, the system sets off a personalized yoga recommendation engine that maps detected autonomic states to evidence-based asanas from the Common Yoga Protocol (CYP) to effectively restore emotional wellness. The proposed system aims to resolve the stress and provide personalized behavioral recovery to enhance student well-being. Keywords: Artificial Intelligence (AI); Internet of Things (IoT); Wearable Sensors; Explainable Artificial Intelligence (XAI); Student Stress Detection; Personalized Yoga Intervention