Abstract
Background: Language outcomes after speech and language therapy in post-stroke aphasia are challenging to predict. This study examines behavioral language measures and resting state fMRI (rsfMRI) as prognostics for response to language therapy. Methods: Seventy patients with chronic aphasia were recruited and treated for one of three deficits: anomia, agrammatism, or dysgraphia. Treatment effect was measured by performance on a treatment-specific language measure, assessed before and after three months of language therapy. Each patient also underwent an additional 27 language assessments and an fMRI scan at baseline. Patient scans were decomposed into 20 components by group independent component analysis, and each component time series was summarized by its fractional amplitude of low-frequency fluctuations (fALFF). Results: Treatment effects were modelled with elastic net regression, using clinical language measures and fALFF imaging predictors independently. Correlation analyses showed high performance for language measures in anomia (r = 0.958, n = 30) and for fALFF predictors in agrammatism (r = 0.940, n = 11) and dysgraphia (r = 0.925, n = 18). These models are state-of-the-art for aphasia recovery prediction. Conclusion: Predicting aphasia recovery with rsfMRI features may outperform predictions from clinical language measures in some patient populations. This suggests rsfMRI may have prognostic value for chronic aphasia patients undergoing language therapy. Differentiating patients who respond to therapy from those who do not is a first step towards personalized treatment in post-stroke aphasia.