Employee productivity is a critical factor influencing organizational performance, operational efficiency, and overall business success. In modern digital workplaces, productivity loss often occurs gradually and remains unnoticed until it results in delayed deliverables, reduced work quality, or employee burnout. Traditional productivity monitoring systems typically rely on invasive techniques such as keystroke logging, screenshot capture, and application usage tracking, which raise significant privacy concerns and negatively impact employee trust. These systems also focus primarily on activity duration rather than analyzing behavioral patterns that indicate cognitive disengagement or fatigue. This research proposes an Intelligent System for Detecting Productivity Loss in Employees using behavioral analytics and machine learning techniques. The proposed system collects non-intrusive behavioral metadata such as typing rhythm, mouse movement dynamics, and application switching frequency to identify productivity anomalies without accessing sensitive data. A hybrid machine learning approach is employed using Isolation Forest for anomaly detection and Long Short-Term Memory (LSTM) Autoencoder for analyzing temporal behavioral patterns. The system architecture integrates a Spring Boot backend, React-based frontend, and a Python-based machine learning engine using FastAPI. Behavioral data is continuously analyzed to detect deviations from normal productivity patterns and generate real-time alerts through an interactive dashboard. This enables early identification of productivity decline while maintaining user privacy. Experimental evaluation demonstrates that the proposed system effectively detects abnormal work patterns and provides actionable insights for improving employee performance and well-being. The system offers a scalable, privacy-preserving, and intelligent approach to workforce productivity monitoring in modern digital work environments.
IJERST (Sat,) studied this question.