Happy work: Improving enterprise human resource management by predicting workers' stress using deep learning.

Area of Science:

• Occupational Health
• Artificial Intelligence
• Human Resource Management

Background:

• Excessive occupational stress negatively impacts employee productivity, safety, and health.
• Current stress assessment methods, like questionnaires, face challenges in data collection and are prone to subjective bias.
• Effective human resource management (HRM) requires accurate prediction of employee stress and identification of its causes.

Purpose of the Study:

• To develop and validate a Deep Learning (DL) based approach for accurately predicting occupational stress.
• To overcome the limitations of traditional stress assessment methods.
• To provide HRM departments with a tool for proactive stress management.

Main Methods:

• Development of two DL models: a stress classification model and a stress regression model.
• Design of two corresponding neural network architectures.
• Training models using employee data including salary, working hours, and Key Performance Indicators (KPIs).

Main Results:

• The DL-based approach achieved 71.2% accuracy in predicting employee stress status using the classification model.
• The regression model demonstrated a prediction loss of 11.1.
• Validation conducted on two real-world datasets (ESI and HAJP).

Conclusions:

• The proposed DL approach effectively predicts occupational stress status.
• Accurate stress prediction enables improved HRM strategies for employee well-being.
• This method offers a data-driven solution to mitigate the negative effects of workplace stress.