Spatiotemporal Granger causality for developmental effective connectivity in infants

• A novel spatiotemporal Granger causality model was proposed for discovering causal relationships from time series, addressing effective connectivity (EC) in the infant brain during the first year of life.¹
• Infant brain undergoes significant structural and physiological transformations in the first year; most prior work used traditional functional connectivity rather than effective connectivity.¹
• The method differs from existing deep learning-based models that infer connectivity and is directly applicable to neural/physiological time series and developmental brain networks.^{1 1}

Gardner updates

• A spatiotemporal causal model reveals developmental changes in infants’ brain effective connectivity networks during the first year of life using novel spatiotemporal Granger causality from time series. ¹

Weekly enrichment (2026-07-20)

• The study was published in IEEE Transactions on Biomedical Engineering (2025, DOI 10.1109/TBME.2025.3596893), proposing a spatiotemporal Granger causality model to discover causal relationships directly from neural time series.^{2 3}
• Unlike prior deep-learning Granger models that read causality off the first-layer network weights, this model learns spatiotemporal features from the time series and applies a sparsity-inducing penalty to those features to extract Granger causality.²
• On simulated datasets the model reached an average AUROC of 0.879 and AUPRC of 0.828, outperforming baseline methods by up to 5.7% in AUROC and 3.9% in AUPRC.²
• Applied to resting-state EEG from healthy infants, it revealed right-hemispheric lateralization of effective connectivity between 3 and 9 months, shifting toward hemispheric symmetry between 9 and 12 months.²
• The maturation of effective connectivity followed a posterior-to-anterior trajectory, beginning in the occipital region and advancing through parietal, temporal, and frontal regions, consistent with the known myelination sequence.²
• Graph analysis showed that as infants aged their brain networks developed higher clustering coefficients and greater global efficiency alongside reduced shortest path lengths, indicating more integrated, efficient network topology.²
• The authors describe this as the first work to uncover changes in infants’ brain effective-connectivity networks across the first year of life using a causal-discovery approach, rather than the more common undirected functional connectivity.^{2 3}
• Exact infant sample size, age-bin counts, and EEG channel/montage details were not reported in the openly available abstract records.³

Footnotes

1. http://ieeexplore.ieee.org/document/11119780 ↩ ↩² ↩³ ↩⁴ ↩⁵

2. https://doi.org/10.1109/tbme.2025.3596893 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

3. https://ieeexplore.ieee.org/abstract/document/11119780/ ↩ ↩² ↩³