Comparing language outcomes in monolingual and bilingual stroke patients

Area of Science:

• Neuroscience
• Neurolinguistics
• Computational Neuroscience

Background:

• Stroke recovery research primarily focuses on monolingual patients.
• Bilingualism is common globally, yet its impact on post-stroke language prognosis is understudied.
• Existing prognostic models may not generalize to bilingual stroke survivors due to potential differences in neural language networks.

Purpose of the Study:

• To evaluate the applicability of prognostic models developed from monolingual stroke patient data to bilingual stroke patients.
• To investigate whether bilingualism influences language recovery prediction accuracy after stroke.
• To identify differences in lesion-deficit associations between monolingual and bilingual stroke patients.

Main Methods:

• Developed prognostic models using lesion data and language assessments from monolingual stroke patients (n=174).
• Applied these models to predict language outcomes for both monolingual and bilingual stroke patients (n=33).
• Compared prediction errors and used Bayesian analysis to assess lesion-deficit associations across groups.

Main Results:

• Models performed equally well in distinguishing good vs. poor outcomes within each group.
• Models were overly optimistic when predicting bilingual language outcomes, with bilinguals showing poorer skills than predicted in 13/22 language tasks (P < 0.05).
• Both groups showed sensitivity to damage in similar brain regions, but bilinguals exhibited greater sensitivity.

Conclusions:

• Post-stroke prognostic models trained on monolingual data do not fully generalize to bilingual patients.
• Bilingual stroke survivors may experience different language recovery trajectories than monolinguals.
• Further research is needed to develop tailored prognostic tools for bilingual stroke patients, considering distinct neural network involvement.