Evaluating the Effectiveness of a Privacy-First AI-Based Multimodal Mood Monitoring System (Sanrakshak) among Office Employees

Abstract

Employee burnout has emerged as a critical occupational health concern in digitally intensive workplaces, particularly in high-demand sectors such as the Information Technology (IT) industry [1][3]. This study proposes a privacy-preserving multimodal Artificial Intelligence framework for early detection of burnout-related emotional exhaustion using aggregated behavioural and linguistic indicators [7] [11]. Drawing on the multidimensional burnout model, the system integrates sentiment dynamics, workload variability, and interaction-pattern features while avoiding personally identifiable data [5] [11]. Empirical evaluation demonstrates that multimodal mood scores significantly predict emotional exhaustion (R² = 0.48, p < .01), showing an 18 % improvement in predictive sensitivity compared with single-modality approaches [7]. The report outlines the importance of ethical governance, privacy, and transparency in AI-based monitoring [11]. The outcomes show that privacy-first multimodal monitoring can facilitate the early identification of a variety of problems which helps in improving employee privacy and trust [9] [10] [11].