Neural tracking of natural speech: an effective marker for post-stroke aphasia

After a stroke, approximately one-third of patients suffer from aphasia, a language disorder that impairs communication ability. Behavioural tests are the current standard to detect aphasia, but they are time-consuming, have limited ecological validity and require active patient cooperation. To address these limitations, we tested the potential of EEG-based neural envelope tracking of natural speech. The technique investigates the neural response to the temporal envelope of speech, which is critical for speech understanding by encompassing cues for detecting and segmenting linguistic units (e.g. phrases, words and phonemes). We recorded EEG from 26 individuals with aphasia in the chronic phase after stroke (>6 months post-stroke) and 22 healthy controls while they listened to a 25-min story. We quantified neural envelope tracking in a broadband frequency range as well as in the delta, theta, alpha, beta and gamma frequency bands using mutual information analyses. Besides group differences in neural tracking measures, we also tested its suitability for detecting aphasia at the individual level using a support vector machine classifier. We further investigated the reliability of neural envelope tracking and the required recording length for accurate aphasia detection. Our results showed that individuals with aphasia had decreased encoding of the envelope compared to controls in the broad, delta, theta and gamma bands, which aligns with the assumed role of these bands in auditory and linguistic processing of speech. Neural tracking in these frequency bands effectively captured aphasia at the individual level, with a classification accuracy of 83.33% and an area under the curve of 89.16%. Moreover, we demonstrated that high-accuracy detection of aphasia can be achieved in a time-efficient (5-7 min) and highly reliable manner (split-half reliability correlations between R = 0.61 and R = 0.96 across frequency bands). In this study, we identified specific neural response characteristics to natural speech that are impaired in individuals with aphasia, holding promise as a potential biomarker for the condition. Furthermore, we demonstrate that the neural tracking technique can discriminate aphasia from healthy controls at the individual level with high accuracy, and in a reliable and time-efficient manner. Our findings represent a significant advance towards more automated, objective and ecologically valid assessments of language impairments in aphasia.