Integration of tongue image features and tongue coating microbiome for differentiating dampness patterns in MASLD

Metabolic dysfunction-associated steatotic liver disease (MASLD) is a prevalent metabolic disorder that can progress to hepatic fibrosis, cirrhosis, and hepatocellular carcinoma. Traditional Chinese medicine (TCM) has therapeutic effects in MASLD. In TCM, dampness patterns predominate in MASLD and are further classified into damp heat (DH) and cold dampness (CD) subtypes based on distinct pathogenic mechanisms. However, the characteristics of the tongue image and tongue coating microbiome across different dampness patterns and their associations with clinical features remain poorly understood.

This study enrolled 320 participants, including 214s dampness patterns patients with MASLD, further classified into DH pattern (n = 110) and CD pattern(n = 104), along with 106 healthy controls. Tongue images and tongue coating samples were collected; tongue image segmentation was performed using a customized Tongue-InSPyReNet framework and extract quantitative features, while tongue coating microbiota were profiled using 16S rRNA gene sequencing. To distinguish between patterns, five machine learning models were developed and evaluated based on integrated tongue image features and microbiome data.

Significant differences in tongue features and microbial composition were observed between groups. The DH pattern was characterized by a red-crimson tongue with yellow coating and enrichment of Prevotella (P < 0.05). In contrast, the CD pattern more frequently exhibited a pale tongue with petechiae and higher abundances of Streptococcus and Rothia (P < 0.05). Integration of tongue image features and tongue coating microbiome effectively distinguished DH and CD patterns, achieving an AUC of 0.871 and an accuracy of 79.1%.

Our study highlights the contributions of tongue image features and the tongue coating microbiome to differentiating two TCM patterns in MASLD, and may provide the rationale for adopting different treatment strategies for different TCM syndromes of MASLD in the future.