Fig. 2

Global latent tree model for Traditional Chinese Medicine clinical features of functional dyspepsia constructed using the overall sample dataset (n = 400). Latent tree model is an undirected tree with the observed variables located at the leaf nodes and the latent variables at the internal nodes. It explains the relationships between the observed variables (i.e., clinical features) and their latent variables using conditional probability distributions. “Y”s are the latent variables in the latent tree model. The number in parentheses is the number of clusters in the latent variables from the probabilistic partition of participants. All latent variables in the above model contained two participant clusters. Based on the manifestation of probabilistic co-occurring clinical features, one of the clusters included participants that were classified into that latent variable, while the other included those that were not classified into that latent variable