Hybrid Modeling of Whole-Brain Dynamics balances Data Fidelity and Interpretability

Lopez Montero D, Liverani L, Zuazua E, Kobeleva X (2026)

Publication Language: English

Publication Status: Submitted

Publication Type: Unpublished / Preprint

Future Publication Type: Article in Edited Volumes

Publication year: 2026

Open Access Link: https://dcn.nat.fau.eu/wp-content/uploads/lopezMliveraniZuazuaKobeleva2026neurosciMod.pdf

Abstract

Whole-brain models (WBMs) are crucial for understanding large-scale neural dynamics. Two main approaches exist: (i) physics-based WBMs built on coupled oscillators such as the Hopf oscillator, which rely on rigid parametrizations with few parameters and computationally expensive model fitting; and (ii) data-driven models such as Neural SDEs, which offer scalable learning but are sample-inefficient and lack interpretable structure. We propose a hybrid modeling framework combining both models and using gradient descent-based fitting to simultaneously optimize multiple metrics representing neural activity, such as network structure and its dynamics. We demonstrate that hybrid models with a data-driven connectome component achieve higher data fidelity than both physics-based WBMs and data-driven models across static and dynamic network descriptors, while preserving interpretable oscillator based structure and replacing the need for exhaustive parameter searches. These results highlight the flexibility and scalability of hybrid WBMs and support their utility for personalized modeling and future digital brain twin applications.

Authors with CRIS profile

Daniel Lopez Montero Lehrstuhl für Dynamics, Control, Machine Learning and Numerics (Alexander von Humboldt-Professur) Lorenzo Liverani Lehrstuhl für Dynamics, Control, Machine Learning and Numerics (Alexander von Humboldt-Professur) Enrique Zuazua Iriondo Lehrstuhl für Dynamics, Control, Machine Learning and Numerics (Alexander von Humboldt-Professur)

Involved external institutions

Universitätsklinikum Bonn DE Germany (DE)

How to cite

APA:

Lopez Montero, D., Liverani, L., Zuazua, E., & Kobeleva, X. (2026). Hybrid Modeling of Whole-Brain Dynamics balances Data Fidelity and Interpretability. (Unpublished, Submitted).

MLA:

Lopez Montero, Daniel, et al. Hybrid Modeling of Whole-Brain Dynamics balances Data Fidelity and Interpretability. Unpublished, Submitted. 2026.

BibTeX: Download