Disconnection-Based Prediction of Poststroke Dysphagia

Area of Science:

• Neuroscience
• Neurology
• Medical Imaging

Background:

• Dysphagia is a frequent and serious complication following a stroke.
• The precise brain regions responsible for swallowing control after stroke remain unclear.
• Existing lesion studies may not fully account for structural disconnections contributing to dysphagia.

Purpose of the Study:

• To predict and explain the link between post-stroke dysphagia and the topological distribution of structural brain disconnections.
• To utilize a multivariate predictive framework to analyze these relationships.

Main Methods:

• Enrolled ischemic stroke patients categorized by nutritional status (oral vs. non-oral).
• Employed propensity score matching to create comparable groups.
• Estimated the structural disconnectome using lesions segmented from acute diffusion-weighted MRI.
• Developed a predictive model incorporating the structural disconnectome and demographic data.

Main Results:

• A disconnectome-based model accurately predicted post-stroke dysphagia (68.6% accuracy).
• Combining disconnectome data with demographics improved prediction accuracy to 72.1%.
• Key predictive connections involved the right insula and left motor strip.

Conclusions:

• Structural disconnectome analysis from acute MRI can predict post-stroke dysphagia.
• Direct and indirect disconnections within the motor network are highly informative for dysphagia prediction.