Estimation and validation of individualized dynamic brain models with resting state fMRI

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Developing generative, causal models of the human nervous system is a key challenge.
• Existing models are either biologically plausible but not individualized, or data-driven but not mechanistic.
• A gap exists in creating individualized, mechanistic models of brain activity.

Purpose of the Study:

• To bridge the gap between biologically plausible and data-driven models.
• To introduce a new modeling approach for individualized neuroscience.
• To develop a method for creating mechanistic, data-driven brain models.

Main Methods:

• Developed Mesoscale Individualized Neurodynamic (MINDy) modeling.
• Fit nonlinear dynamical systems models directly to human brain imaging data.
• Generated data-driven network models for thousands of interacting brain regions rapidly.

Main Results:

• MINDy models were demonstrated to be valid, reliable, and robust.
• MINDy models accurately predicted individualized patterns of resting-state brain activity.
• MINDy outperformed functional connectivity methods in uncovering mechanisms of individual differences.

Conclusions:

• MINDy modeling provides a novel, efficient, and robust approach for creating individualized, mechanistic brain models.
• This framework advances the understanding of individual differences in brain dynamics.
• MINDy modeling offers a powerful tool for personalized neuroscience research.