This study examines factors influencing occupational injuries among plant and machine operators using the Semi-supervised MarginBoost algorithm. Data from the 2007-2009 Korean National Health and Nutrition Examination Survey (KNHANES) were analyzed, covering 4,062 employed participants. The MarginBoost model achieved 84.3% accuracy, outperforming other models. Key factors identified included exposure to hazardous substances, ergonomic conditions, and psychosocial stress. The findings emphasize the need for targeted interventions to enhance workplace safety and offer a robust predictive tool for the effective management of occupational health.

Abstract
1. Introduction
2. Materials and Methods
    2.1. Study Design and Population
    2.2. Data Collection
    2.3. Input Variables
    2.4. Data Preprocessing
    2.5. Machine Learning Models
    2.6. MarginBoost Algorithm
    2.7. Model Training and Evaluation
    2.8. Feature Importance
3. Results
    3.1 Health-Related Characteristics
    3.2. Work Environment Characteristics
    3.3. Occupational Injury Rates
    3.4. Model Performance
    3.5. Feature Importance
4. Discussion
References

• 변해원(Department of AI-Software, Inje University, South Korea) | Haewon Byeon Corresponding author