HR Challenges in Managing Hybrid and Remote Work Forces Using Deep Learning: An Implementation Framework

Abstract

Hybrid and remote Work forces are the new standard, which means human resource (HR) management needs extraordinary tools to ensure the well-being, productivity, and engagement of personnel. We present the development and evaluation of a deep learning model for automated detection of employee sentiment, burnout risk, and productivity anomalies, on multimodal data obtained from corporate communications and HR data systems in an organization. We used a 5,000-staff dataset from a large open-source hybrid workforce benchmark (the ‘HybridWorkWellness' dataset) over 12 months to validate the approach. The sentiment classifier obtained 0.91 Fl-score, burnout prediction obtained 0.87 AUC and anomaly detection had 0.92 precision. Other statistics were recall 0.88, accuracy 0.93, specificity 0.89, sensitivity 0.91, Matthews correlation coefficient 0.86. Results indicate that the framework can significantly reduce manual HR interventions by 34%, while enhancing risk detection for distributed teams.

Ähnliche Arbeiten