Abstract:Many functional magnetic resonance imaging (fMRI) studies rely on estimates of hierarchically organised brain networks whose segregation and integration reflect the dynamic transitions of latent cognitive states. However, most existing methods for estimating the community structure of networks from both individual and group-level analysis neglect the variability between subjects and lack validation. In this paper, we develop a new multilayer community detection method based on Bayesian latent block modelling. The method can robustly detect the group-level community structure of weighted functional networks that give rise to hidden brain states with an unknown number of communities and retain the variability of individual networks. For validation, we propose a new community structure-based multivariate Gaussian generative model to simulate synthetic signal. Our result shows that the inferred community memberships using hierarchical Bayesian analysis are consistent with the predefined node labels in the generative model. The method is also tested using real working memory task-fMRI data of 100 unrelated healthy subjects from the Human Connectome Project. The results show distinctive community structure patterns between 2-back, 0-back, and fixation conditions, which may reflect cognitive and behavioural states under working memory task conditions.

What problem does this paper attempt to address?

### Problems the paper attempts to solve This paper aims to solve two main problems in the estimation of brain network community structure in functional magnetic resonance imaging (fMRI) research: 1. **Inter - individual variability**: Most of the existing methods ignore the variability between individuals when analyzing the community structure of networks at the individual and group levels. This variability is crucial for understanding the differences in the performance of different individuals in cognitive tasks. 2. **Insufficient method validation**: Existing methods lack effective validation means, and it is difficult to evaluate their performance on real - data. To this end, the authors developed a new multi - layer community detection method based on the Bayesian Latent Block Model (LBM). This method can robustly detect the group - level community structure of weighted functional networks and preserve the variability of individual networks. In addition, the authors also proposed a new multivariate Gaussian generative model based on the community structure to simulate synthetic signals to verify the effectiveness of the method. ### Specific objectives - **Robust detection of community structure**: Through the Bayesian Latent Block Model and Markov Chain Monte Carlo (MCMC) sampling, robustly detect the group - level community structure of weighted functional networks while considering an unknown number of communities. - **Preserve individual variability**: While detecting the group - level community structure, preserve the variability of individual networks, thereby more accurately reflecting the differences between individuals. - **Method validation**: Use the multivariate Gaussian generative model to simulate synthetic data to verify the performance of the method. The results show that the inferred community membership is consistent with the predefined node labels in the generative model. - **Apply to real - data**: Apply the method to the fMRI data of memory tasks of 100 healthy subjects from the Human Connectome Project, showing different community structure patterns under 2 - back, 0 - back and fixed conditions, which may reflect the cognitive and behavioral states under working memory task conditions. ### Method overview 1. **Individual - level modeling**: - Use the Latent Block Model (LBM) to model the weighted adjacency matrix of each subject. - Estimate the individual - level community membership through MCMC sampling. - The Latent Block Model parameters are assumed to independently follow the conjugate Normal - Inverse - Gamma (NIG) prior distribution. 2. **Group - level modeling**: - Use the Categorical - Dirichlet conjugate pair to model the estimated individual - level community membership. - Define the Maximum Label Assignment Probability Matrix (MLAPM) to provide information on the group - level community structure. - Use the Normal - NIG conjugate pair to model the individual - level functional connections and estimate the mean and variance connections of the group - representative network. ### Experimental verification 1. **Generative model and synthetic data**: - Use the multivariate Gaussian generative model to simulate synthetic data to verify the performance of the method. - The results show that the MLAPM estimates are consistent with the predefined latent label vectors and the number of communities in the generative model. 2. **Task - fMRI data experiment**: - Apply the method to the fMRI data of memory tasks of 100 healthy subjects. - Show different community structure patterns under 2 - back, 0 - back and fixed conditions, reflecting the changes in cognitive and behavioral states. Through these methods and experiments, the paper has successfully solved the problem of inter - individual variability in the estimation of functional brain network community structure and provided effective method validation means.

Hierarchical Bayesian inference for community detection and connectivity of functional brain networks