Academic

Functional connectivity represents brain network interactions and is fundamental to the translation of neural structure to brain function. While multiple approaches have been proposed for mapping functional connectivity based on statistical associations between neural activity, association does not necessarily incorporate causation. Additional approaches have been proposed to incorporate aspects of causality to turn functional connectomes into causal functional connectomes, however they focus on specific aspects of causality. This warrants a systematic statistical framework to causal functional connectomics to evaluate existing approaches and guide the development of further causal methodologies. In this work, we first establish such a statistical guide. We particularly focus on the introduction of directed graphical models as a framework, which defines the directed Markov property as an essential criterion for capturing causality in the proposed functional connectomes. Based on these notions, we perform a comparative study of existing approaches for inferring causal functional connectivity from neural time series. However, the common formulation of directed graphical modeling is not ideal for neural time series since it was developed for variables with independent and identically distributed samples. Therefore, we develop a novel methodology, coined the Time-aware PC (TPC) algorithm, that adapts directed graphical modeling to the time series scenario. We establish the mathematical guarantee of the TPC algorithm in inferring causal relationships from time series data under standard time series conditions. We then demonstrate the utility of the methodology in simulated and public benchmark datasets, and recent Neuropixels recordings from the mouse visual cortex under different visual stimuli. Lastly, we compute the causal functional connectivity in the human domain in Alzheimer’s disease from resting-state functional magnetic resonance imaging (fMRI) data and perform an exploratory analysis of alteration of causal functional connectivity edges between subjects with Alzheimer’s disease compared to cognitively normal subjects. The corresponding brain regions are found to be in agreement with medical literature on brain regions impacted by Alzheimer’s disease.