Answers are grounded in this paper and cite their sources.
Background Liver cirrhosis is a common end-stage of chronic liver diseases, with high morbidity and mortality. Aims This study aimed to develop a tongue diagnosis model using machine learning and neural networks to evaluate the feasibility of predicting cirrhosis risk from tongue appearance data. Methods Tongue appearance and laboratory data were collected from 115 cirrhosis patients (July 2023–February 2024). MELD scores were calculated. After preprocessing, both image-based and text-based classifiers were constructed using five algorithms: Naive Bayes (NB), Random Forest (RF), Logistic Regression (LR), Support Vector Machine (SVM), and Multilayer Perceptron (MLP). Model performance was assessed using accuracy, F1-score, recall, precision, and AUC (Area Under the ROC Curve). A MELD-only baseline model was also developed for comparison. Results Multi-feature models consistently outperformed the MELD-only baseline. In image-based classification, SVM achieved the highest accuracy (85.71%) with 64-dimensional features. In text-based classification, RF, LR, SVM, and MLP all achieved 82.86% accuracy, demonstrating stable performance. Conclusion Machine learning models incorporating multiple tongue appearance features effectively addressed the limitations of MELD score alone. The SVM model with 64-dimensional image features achieved the best predictive performance, confirming the feasibility of cirrhosis risk prediction through tongue image analysis. This study offers a novel approach for non-invasive clinical risk stratification in cirrhosis patients.
No takes yet. Share an insight, caveat, or question.
Yang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: