Study on the Correlation Between Machine Learning-Based Tongue Features of Healthy Individuals in the Real World and Age and Gender
Objective Validate the correlation between tongue features and age and gender by utilizing real-world tongue image data from healthy individuals.Methods Establishing,training,and validating a model using artificial intelligence image acquisition techniques and machine learning tools.Results ①Artificial intelligence has achieved the digitization and standardization of Traditional Chinese Medicine tongue diagnosis data,establishing a database of Traditional Chinese Medicine tongue diagnosis(n=56573).②A highly accurate random forest model has demonstrated the correlation between tongue features and gender and age.③Further selection of key factors that show differences in tongue features between different age groups and genders.④Machine learning artificial neural network models were trained,tested,and validated(training set R2=0.8772,validation set R2=0.8715,test set R2=0.8707,AUC=89.5％),demonstrating excellent accuracy.Conclusion Machine learning validation using real-world data has confirmed that tongue features change with age.Additionally,analysis of large-scale real-world data has revealed that there are different tongue features associated with gender differences.
Real-world dataMachine learningTongue image and age relationshipTongue image and gender relationship
万方数据
