Objective: Primary palmar hyperhidrosis (PPH), characterized by excessive palm sweating, significantly impacts patients’ physiology, psychology, self-esteem, work, life, and social interactions. The incidence of depression is higher among PPH patients. Timely detection of key predictive factors and the development of risk prediction models are crucial for effective intervention and treatment in this patient group.

Design: We conducted an in-depth analysis of clinical data from 926 PPH patients treated at the Thoracic Surgery Department of Beijing Haidian Hospital between 2016 and 2021. We utilized the Boruta algorithm alongside the Backward Elimination strategy to select predictive factors and constructed five machine learning models. By evaluating these models’ performance, we determined the optimal one. Additionally, we introduced the Shapley Additive exPlanations (SHAP) method to enhance the interpretability of this optimal model.

Results: The personality diagnostic questionnaire-4 (PDQ-4) score, self-rating anxiety scale (SAS) score, family history, quality of life excluding PPH (QoL-ex), onset age, and the age when PPH begins to impact life (Impact age) are six predictive factors for depression in PPH patients. The support vector machine (SVM) model performs more comprehensively through model validation. In the validation set, the area under the curve (AUC) is 0.798 (95% CI: 0.737−0.858), with a Brier score of 0.1451 (95% CI: 0.1233−0.1716), accuracy of 0.7184, sensitivity of 0.775, specificity of 0.699, and F1 score of 0.585.

Conclusion: These findings can enhance our understanding of depression in PPH patients, and the SVM model is a valuable screening tool for assessing the risk of depression in PPH patients.

Xiaoyu Wang has completed his PhD from Tianjing Medical University. He is the attending physician of Department of Thoracic Surgery, Haidian Section of Peking University Third Hospital. He has conducted in-depth research on the psychological state of patients with hyperhidrosis and published relevant articles in well-known journals