J Thorac Dis. 2026 Feb 28;18(2):146. doi: 10.21037/jtd-2025-aw-2074. Epub 2026 Feb 25.
ABSTRACT
BACKGROUND: Compensatory hyperhidrosis (CH) remains the most prevalent postoperative adverse event following endoscopic thoracic sympathectomy (ETS) for primary hyperhidrosis (PH). Current predictive models lack reliability in estimating CH severity. This study introduces a novel predictive framework utilizing rough set theory to establish decision rules for CH stratification.
METHODS: In this single‑center retrospective cohort study, clinical data from 225 PH patients undergoing ETS were analyzed, including 37 predictive indicators. These variables were subjected to correlation analysis, regression analysis, and rough set analysis with CH severity.
RESULTS: There were 93.3% (210/225) of patients exhibiting CH following ETS, with 33.3% classified as grade III CH, and no grade IV CH was noted. Body mass index (BMI), the level of sympathectomy, and the temperature difference of the right hand after surgery and before anaesthesia were shown to be significantly correlated with CH on correlation analysis. However, no valid regression model was established with significant correlations involving indicators for further regression analysis. By switching to rough set analysis, four predictive rules for grade III CH were derived: (I) BMI >22 kg/m2 + initial onset age of PH >11 years, 84% accuracy; (II) BMI 19.5-22 kg/m2 + surgical age >28.5 years, 82% accuracy; (III) BMI 18.5-19.4 kg/m2 + postoperative right-hand temperature >36.6 ℃, 77% accuracy; (IV) BMI <18.5 kg/m2 + postoperative right-hand temperature <37.0 ℃ + initial PH onset age <10 years, 71% accuracy.
CONCLUSIONS: Rough set analysis provides a promising approach for exploring the patterns of CH severity following ETS in patients with PH, and thus which merits further investigation through multicenter, large-sample studies. The four preliminary decision rules for predicting grade III CH derived from rough set analysis show potential clinical relevance but remain tentative, as their utility requires validation in prospective cohorts prior to widespread clinical application.
PMID:41816421 | PMC:PMC12972784 | DOI:10.21037/jtd-2025-aw-2074
