Using machine learning algorithm to predict the risk of post-traumatic stress disorder among firefighters in Changsha

Area of Science:

• Occupational Health
• Psychological Medicine
• Data Science

Background:

• Firefighters face significant psychological trauma and a high risk of developing post-traumatic stress disorder (PTSD).
• Early identification and intervention are crucial for improving prognosis in firefighters with PTSD, yet reliable predictive models are lacking.

Purpose of the Study:

• To develop and validate a machine learning algorithm for accurately predicting PTSD onset in firefighters.
• To identify key psychological traits and work-related factors that serve as significant predictors of PTSD risk.

Main Methods:

• A cross-sectional survey was conducted with 628 firefighters from 20 fire brigades in Changsha.
• Data preprocessing involved the synthetic minority oversampling technique (SMOTE) and parameter tuning via grid search.
• The predictive performance of machine learning models was evaluated using 5-fold cross-validation, ROC-AUC, accuracy, precision, recall, and F1 score.

Main Results:

• The random forest model demonstrated strong predictive capability, achieving an average ROC-AUC of 0.790, 90.1% accuracy, and an F1 score of 0.945.
• Key predictors of PTSD onset included perseverance (0.165), forced thinking (0.158), and reflective deep thinking (0.152).
• Other significant predictors identified were employment time, psychological power, and optimism.

Conclusions:

• A random forest-based PTSD onset prediction model for firefighters exhibits robust predictive power.
• Psychological characteristics and work situations are valuable predictors for assessing PTSD risk in firefighters.
• Further validation with larger datasets is recommended to ensure clinical applicability of the predictive models.