Multimodal Deep Learning Framework for Employee Stress Detection in the Workplace

Workplace stress is a pervasive issue impacting employee well-being and organizational productivity. This study proposes a multimodal AI framework for detecting stress in employees using structured workplace data, such as age, job role, workload, overtime, and support metrics. Machine learning (ML) algorithms, including Decision Tree, Random Forest, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Na¨ıve Bayes, were benchmarked against deep learning (DL) models, such as Convolutional Neural Networks (CNN), Multilayer Per- ceptron (MLP), Bidirectional Long Short-Term Memory (BiL- STM), and Attention-BiLSTM. Experimental results demonstrate exceptional performance, with Decision Tree and Random Forest achieving accuracies of 99.83% and 99.67%, respectively, and CNN and BiLSTM reaching 99.00% and 98.50%. Explainable AI (XAI) techniques, including SHAP and LIME, were integrated to enhance model interpretability, ensuring actionable insights for HR teams. The framework offers a scalable, ethical, and deploy- able solution for proactive stress management, with potential for real-time monitoring using physiological and behavioral data.