[Predictive Capacity of a Clinical Questionnaire to Identify the Propensity for Occupational Mental Health Disorders]

The process of qualifying work-related mental health disorders in Chile is regulated by Law No. 16.744. However, it is frequently challenged in terms of precision and objectivity.

To assess the feasibility of machine learning algorithms in supporting mental health qualification.

Application of a mental health disease propensity questionnaire to a sample of 340 Chilean workers. Evaluation of decision tree models and logistic regression, including enhanced versions with additional variables ("Plus") related to the intensity and temporality of stressors.

The "Plus Tree" model showed higher accuracy (91.2%, 95% CI: 83.9%-95.9%), particularly with balanced data (76.9%). However, there is room for improving the model's specificity due to data imbalance caused by the presence or absence of mental health condition.

The inclusion of algorithms in the qualification process enhances efficiency and objectivity. However, further database expansion and model refinement are necessary to achieve better precision.