Digital Twin Brain: Generating Multitask Behavior from Connectomes for Personalized Therapy.

Area of Science:

• Neuroscience
• Computational Psychiatry
• Digital Twins

Background:

• Personalized psychiatry necessitates digital twin models for predicting diverse functions from individual neurobiology.
• Current models face challenges in connecting brain structure to complex, multitask behavior, limiting clinical application.

Purpose of the Study:

• To introduce and validate a digital twin brain framework.
• To translate individual brain connectomes into predictions of neurobehavioral dynamics.
• To enable personalized functional modulations and interventions.

Main Methods:

• A novel 2-component architecture: a hypernetwork personalizing a main network using an individual's connectome.
• A recurrent neural network simulates participant-specific behavioral and blood-oxygen-level-dependent (BOLD) time series.
• Gradient backpropagation identifies connectome manipulations for targeted functional modulation.

Main Results:

• The model achieved >90% accuracy in predicting behavioral choices and high fidelity for reaction times (r > 0.85) and BOLD patterns (r = 0.84).
• In silico interventions successfully modulated targeted functions.
• The framework reproduced realistic interindividual variability in treatment effects based on baseline connectomes.

Conclusions:

• The digital twin brain system facilitates high-fidelity in silico prediction of complex neurobehavioral functions.
• It enables personalized modulation of these functions, advancing individualized psychiatric care.
• This approach links brain connectomes directly to behavior, creating a mechanistic platform for intervention design.