Suicide probability among physicians: an explainable machine learning analysis of depression, burnout, anxiety, and coping styles

The study examined suicide risk among 769 physicians in Turkey using questionnaires assessing depression, anxiety, burnout, and coping strategies. The research found that depressive symptoms are the most significant factor influencing suicide probability, followed by anxiety, burnout, and negative coping mechanisms. Researchers employed two analytical methods: conventional linear regression and advanced machine learning model XGBoost. Both methods produced similar results, confirming the consistency of findings. However, the XGBoost model did not provide a substantial predictive advantage over traditional regression. The authors emphasize that these findings should be considered as hypotheses for further research and not as a validated tool for clinical prediction without additional external validation.