Lesion-Derived Estimated Functional and Structural Connectomes Predict Post-Stroke Recovery Outcomes

Lesion-derived estimated structural disconnection (eSDC) and functional connectivity (eFC) from N=96 stroke patients predict post-stroke motor, language, attention, and memory outcomes at multiple timepoints, outperforming lesion-only and observed-FC models.

• Why it matters: Prognosis after stroke is difficult, and advanced connectome MRI is often impractical, leaving a gap for accessible network-based predictors of individual recovery.
• What they did: The pipeline uses NeMo to estimate eSDC from lesion masks and Krakencoder to estimate eFC from NeMo outputs, then trains models to predict motor, language, attention, and memory outcomes across timepoints.
• The result: eSDC/eFC models predicted almost all outcomes—especially motor—retain inter-individual variability versus observed functional connectomes, mirror population-level stroke-related functional changes, and yield modality- and domain-specific feature signatures.

AI summary of the abstract · may contain errors · not medical advice · read the paper.