Statistics, Data Science, and the Connectome

Statistics and data science are central to brain connectivity research, enabling extraction of meaningful patterns from complex neuroimaging data and advancing understanding of functional architecture. Connectivity analyses map the intricate networks underlying cognition and disease by revealing how brain regions communicate, synchronize, and influence one another. This article reviews statistical analysis of functional connectivity (i.e., the undirected association between time series from distinct brain regions) and effective connectivity (i.e., the directed causal influence that one region exerts over another). Functional connectivity is often assessed using correlation, coherence, or mutual information, while effective connectivity requires more sophisticated modeling approaches such as Granger causality, dynamic causal modeling, or structural equation models. We highlight key analytic choices in connectivity studies (e.g., preprocessing, parcellation, spatial and temporal resolution across fMRI and EEG), and then describe the metrics and models used to assess functional and effective connectivity. This is followed by recent advances on time-varying, connectivity, high-dimensional graphical networks, and graph theory. We conclude with open problems in brain connectivity research, emphasizing the need for continued statistical innovation and interdisciplinary collaboration.