What problem does this paper attempt to address?

The problem that this paper attempts to solve is to model the network connection patterns in resting - state functional magnetic resonance imaging (rs - fMRI) data by establishing a Hierarchical Bayesian Recurrent State Space Model. Specifically, the author aims to reveal the network patterns shared under different disease conditions, and pays special attention to the changes in neural circuits in patients with mild cognitive impairment (MCI). ### Objectives of the Paper 1. **Modeling Shared and Specific Network Patterns** - This model can simultaneously capture the network patterns shared between healthy individuals and disease groups, as well as the network patterns that are dominant only in a specific group. - By introducing Class Conditional Transition Matrices, the model can distinguish the brain dynamic changes under different health states. 2. **Improving Existing Methods** - The existing standard state - space models have two main limitations: one is that they cannot model the shared connection patterns between healthy and disease groups; the other is that they do not take into account the recurrent state patterns observed by Vidaurre et al. (2017). - The model proposed in this paper overcomes these limitations by introducing recurrent state dependence and Class Conditional Transition Matrices. 3. **Evaluating Model Performance** - The author evaluated the performance of the model using the ADNI2 dataset, especially the ability to identify the latent states in MCI patients. - The results show that the performance of this model on the ADNI2 dataset is better than that of the current state - of - the - art deep - learning methods. ### Overview of the Method - **Model Structure** - Use a Hierarchical Bayesian Recurrent State Space Model, in which the state \( z_t \) at each time point depends not only on the state \( z_{t - 1} \) at the previous time point, but also on the observed value \( x_{t - 1} \) at the previous time point. - Introduce Class Conditional Transition Matrices \( \Pi_c \) to model the brain dynamic changes under different health states, while sharing the observation parameters \( \theta \) and the transformation function \( G \). - **Inference Process** - Use the extended EM algorithm for parameter estimation, and calculate the class - conditional posterior probabilities through the message - passing algorithm. - In the optimization process, first marginalize the hidden states, and then update the shared parameters and class - specific parameters. ### Experimental Results - **Performance Comparison** - On the ADNI2 dataset, this model is superior to the existing deep auto - encoder HMM method in terms of accuracy, sensitivity, specificity, positive predictive value, and negative predictive value. - **Findings** - Healthy individuals show stronger inter - hemispheric cortical connections in some states, especially in the occipital lobe area. - These findings are consistent with the previous results of voxel - mirror - homotopic - connectivity analysis in MCI patients. ### Conclusion This study proposes a new Hierarchical Bayesian Recurrent State Space Model, which can effectively model the network connection patterns in resting - state functional magnetic resonance imaging data, especially in patients with mild cognitive impairment. Future work will be extended to other neuropathological datasets, and more flexible non - linear likelihood models and additional prior structures will be explored.

Bayesian recurrent state space model for rs-fMRI

Modeling Brain Functional Dynamics Via Hidden Markov Models

State-space model with deep learning for functional dynamics estimation in resting-state fMRI

Identification of brain states, transitions, and communities using functional MRI

Bayesian modelling of effective and functional brain connectivity using hierarchical vector autoregressions

Bayesian Models of Functional Connectomics and Behavior

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

A nonstationary nonparametric Bayesian approach to dynamically modeling effective connectivity in functional magnetic resonance imaging experiments

Bayesian Modeling of Effective and Functional Brain Connectivity using Hierarchical Vector Autoregressions

Hierarchical Spatio-Temporal State-Space Modeling for fMRI Analysis

Bayesian Analysis of fMRI data with Spatially-Varying Autoregressive Orders

NHSMM-MAR-sdNC: A novel data-driven computational framework for state-dependent effective connectivity analysis

MCI Detection using fMRI time series embeddings of Recurrence plots

Bayesian Bi-clustering of Neural Spiking Activity with Latent Structures

Bayesian Modeling of the Structural Connectome for Studying Alzheimer Disease

Full Bayesian Modeling for fMRI Group Analysis

Bayesian Joint Modeling of Multiple Brain Functional Networks

A Bayesian General Linear Modeling Approach to Cortical Surface fMRI Data Analysis

Recognizing Brain States Using Deep Sparse Recurrent Neural Network

Hierarchical Bayesian Modeling of Inter-Trial Variability and Variational Bayesian Learning of Common Spatial Patterns from Multichannel EEG

Large-scale Directional Connections among Multi Resting-State Neural Networks in Human Brain: A Functional MRI and Bayesian Network Modeling Study