Application of computer-aided tongue image system for severe mycoplasma pneumoniae pneumonia

Abstract

Background: Mycoplasma pneumoniae pneumonia (MPP) represents the predominant form of community-acquired pneumonia in children. Clinical challenges in identifying severe MPP (SMPP) critically threaten pediatric health. Objective: This study aimed to evaluate the application of the computer-aided system for standardizing tongue image characteristics and diagnosing SMPP. Methods: We enrolled hospitalized children with general MPP (GMPP, n = 243) and SMPP (n = 371) between 2023 and 2024. The SMF-III system was employed to quantify tongue image features. Univariate logistic regression analysis was performed to identify key independent risk factors for SMPP, followed by correlation analysis. ROC curve analysis was conducted to assess diagnostic efficacy. Results: Significant differences in tongue features were observed between the GMPP and SMPP groups. SMPP patients predominantly exhibited red/crimson tongue coloration, yellow-white/yellow coatings, thin-greasy/thick coating textures, and reduced or absent moisture with higher total tongue image scores. Logistic regression confirmed the scores, CRP, NLR, IL-6, and IFN-γ as independent risk factors for SMPP. The scores were positively correlated with CRP, NLR, IL-6, and IFN-γ. Notably, combining tongue image scores with CRP enhanced predictive accuracy for SMPP. Conclusion: Tongue image variations reflect pediatric MPP disease progression. The computer-aided tongue diagnostic system provides a rapid, cost-effective, and reliable tool for auxiliary SMPP diagnosis.