We developed MMBeans by integrating multi-state stochastic block models (SBMs) into Bayesian nonparametric clustering. This approach allows state-specific modular structures to be informed during the subtyping process using informative connectivity features that define identified subtypes. We created a variational inference (VI) algorithm and R scripts to implement this model. The details of the model and its implementation can be found in our paper, "Bayesian subtyping for multi-state brain functional connectome with application to preadolescent brain cognition."
The MMBeans_function.R script contains the VI algorithm for the MMBeans model, while simulation_example.R demonstrates how to implement MMBeans.
Tianqi Chen, Chichun Tan, Hongyu Zhao, Todd Constable, Sarah Yip, and Yize Zhao (2023). Bayesian subtyping for multi-state brain functional connectome with application on adolescent brain cognition.