Developing Action Plans Based on Machine Learning Analysis to Prevent Sick Leave in a Manufacturing Plant.

Area of Science:

• Occupational Health
• Machine Learning
• Predictive Analytics

Background:

• Employee sick leave poses significant challenges for productivity and workplace well-being.
• Proactive health promotion strategies are crucial for mitigating absenteeism in industrial settings.

Purpose of the Study:

• To develop actionable health promotion plans for employees.
• To utilize a machine learning model for predicting sick leave at a Japanese manufacturing plant.

Main Methods:

• A random forest machine learning model was employed to predict sick leave.
• Variable importance and partial dependence plots were used to inform health promotion strategies.

Main Results:

• The predictive model achieved an area under the curve of 0.882.
• Key predictors for sick leave due to mental disorders included high stress response scores (Brief Job Stress Questionnaire), younger age, and specific departments.
• Action plans focused on optimizing the use of the Brief Job Stress Questionnaire and enhancing support for younger employees and managers in high-risk departments.

Conclusions:

• A novel process for developing targeted employee health promotion action plans was established.
• The machine learning approach offers a valuable tool for occupational health practitioners to predict and manage sick leave effectively.