Virtual Brain Inference (VBI)

Virtual Brain Inference (VBI) combines computational models of brain dynamics with individual brain imaging data to enable efficient Bayesian inference on whole-brain models.

Key Features:

• Fast simulations: VBI provides rapid simulation capabilities for whole-brain network models.
• Taxonomy of feature extraction: VBI includes organized feature extraction methods for neuroimaging data analysis.
• Efficient data storage and loading: VBI implements optimized data management for handling large-scale neuroimaging datasets.
• Probabilistic machine learning algorithms: VBI employs Bayesian inference methods for biophysically interpretable parameter estimation from non-invasive and invasive recordings.
• Uncertainty quantification: VBI enables hypothesis evaluation through probabilistic uncertainty estimates in network neuroscience.

Scientific Applications:

• Whole-brain network modeling: VBI advances understanding of complex brain dynamics and neurobiological underpinnings by inverting virtual brain models.
• Precision medicine: VBI enhances the predictive power of virtual brain models for clinical applications.
• In-silico validation: VBI demonstrates accuracy and reliability of inference for commonly used whole-brain network models and their associated neuroimaging data.

Methodology:

VBI performs automated model inversion to estimate control (bifurcation) parameters at large scales associated with neuroimaging modalities of varying spatio-temporal resolutions.