Learning Time-Varying Multi-Region Brain Communications via Scalable Markovian Gaussian Processes

Published: 01 May 2025, Last Modified: 23 Jul 2025ICML 2025 oralEveryoneRevisionsBibTeXCC BY 4.0
TL;DR: We developed a scalable method using Markovian Gaussian Processes (State Space Model) to track how multiple brain regions communicate dynamically over time.
Abstract: Understanding and constructing brain communications that capture dynamic communications across multiple regions is fundamental to modern system neuroscience, yet current methods struggle to find time-varying region-level communications or scale to large neural datasets with long recording durations. We present a novel framework using Markovian Gaussian Processes to learn brain communications with time-varying temporal delays from multi-region neural recordings, named Adaptive Delay Model (ADM). Our method combines Gaussian Processes with State Space Models and employs parallel scan inference algorithms, enabling efficient scaling to large datasets while identifying concurrent communication patterns that evolve over time. This time-varying approach captures how brain region interactions shift dynamically during cognitive processes. Validated on synthetic and multi-region neural recordings datasets, our approach discovers both the directionality and temporal dynamics of neural communication. This work advances our understanding of distributed neural computation and provides a scalable tool for analyzing dynamic brain networks.
Lay Summary: Our brains are made up of many regions that constantly exchange information. But this communication isn’t static because it shifts over time, especially during tasks like seeing or thinking. Existing tools often miss these subtle timing changes or are too slow for analyzing large brain datasets. To solve this, we developed a new method called the Adaptive Delay Model (ADM). ADM uses a combination of Gaussian Processes and State Space Models to detect not just who is talking to whom in the brain, but also when they’re doing it, with precise timing that changes over time. It’s also designed to be fast, so it can handle modern, large-scale brain recordings. We applied ADM to both simulated data and real recordings from mice and monkeys and found that it revealed dynamic patterns of brain communication that match known brain hierarchies. This opens the door to deeper understanding of how brain regions coordinate during perception, thought, or even disease, and could eventually help improve brain-computer interfaces or treatments for neurological conditions.
Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.
Link To Code: https://github.com/BRAINML-GT/Adaptive-Delay-Model
Primary Area: Applications->Neuroscience, Cognitive Science
Keywords: Multiple Brain Region Communications; Markovian Gaussian Processes; State Space Model
Submission Number: 11292
Loading