AI-Powered Predictive Modelling of Job Stress and its Impact on Workforce Productivity in the Pharmaceutical Industry

Abstract

Workforce stress in the pharmaceutical sector is usually overlooked until its influence on output is recognized, owing to its reliance on reactive systems (i.e., surveys, HR employment records, absenteeism). Such traditional methods are disconnected, do not reflect the current reality, and do not provide practical insight. To address these limitations, the proposed system is an AI-powered predictive modeling framework based on multimodal data integration from wearables, behavioral logs, and workplace systems, as well as a processing pipeline involving advanced algorithms (Temporal Convolutional Networks, DeBERTa-v3, and XGBoost with causal estimation). It also anticipates stress levels and measures how they affect productivity, thus avoiding last-minute interventions. The results reveal that the proposed system significantly improves performance, with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 1. 8 \%}$</tex> accuracy on stress prediction, an RMSE of 0.127, and a detection delay of only 3.2 hours, and a 13.4% increase in productivity and a higher level of managerial satisfaction following the intervention underline the proposed system's significance in terms of boosting employee well-being and organizational success.

Ähnliche Arbeiten