Основные понятия
The author introduces the BDyMA method to address challenges in discovering Dynamic Effective Connectome (DEC) by leveraging Bayesian Dynamic DAG learning. The approach enhances accuracy and reliability compared to state-of-the-art methods.
Аннотация
The content discusses the challenges in extracting DEC from brain data, introducing the BDyMA method to improve accuracy and reliability. It highlights the importance of prior knowledge incorporation, simulations on synthetic data, experiments on Human Connectome Project data, and the impact of DTI data. The method's effectiveness is demonstrated through comparisons, reliability tests, sparsity analysis, connectome comparisons, and validity assessments.
- Score-based Directed Acyclic Graph (DAG) discovery methods show improvements in causal structure extraction.
- Challenges faced include high-dimensional dynamic DAG discovery limitations and low-quality fMRI data.
- BDyMA method addresses challenges by incorporating Bayesian Dynamic DAG learning with M-matrices Acyclicity characterization.
- Simulations on synthetic data and experiments on Human Connectome Project data demonstrate improved accuracy and reliability.
- Incorporating DTI data as prior knowledge enhances accuracy and reliability in DEC discovery.
Статистика
Comprehensive simulations on synthetic data and experiments on Human Connectome Project (HCP) data demonstrate that our method can handle both of the two main challenges.
The presented dynamic causal model enables us to discover direct feedback loop edges as well.