Comparison of Neural Tracking and Spectral Entropy in Patients with Disorders of Consciousness

Abstract

ObjectivesThis study aims to explore the brain responses of patients with disorders of consciousness (DoC) to natural speech. Specifically, it focuses on a key characteristic of natural speech: the speech envelope. To achieve this, we employed two distinct measures. The first evaluates how effectively the patients brain activity "tracks" the speech envelope, called neural tracking. The second assesses the complexity of the brains responses to the speech stimulus, measured through spectral entropy. These two measures are then compared in their association with the patients clinical diagnosis and their level of behavioral responsiveness, both of which were assessed using the Coma Recovery Scale-Revised (CRS-R). DesignFour patients with DoC participated in this study, during which their brain activity was recorded using electroen-cephalography (EEG). At the same time, they listened to a narrated story in both Dutch and Swedish. Additionally, EEG baseline recordings were collected. We employed a backward modeling approach to evaluate the speech envelopes neural tracking. This technique involves training a model to map the relationship between EEG signals and the corresponding speech envelope. Once the model is trained, it can use unseen EEG data to reconstruct the speech envelope, which is then compared to the original speech envelope to assess how effectively the patient processed the auditory stimulus. For the behavioral assessment, we recalculated the CRS-R score of each patient into the CSR-R index, a more meaningful score that utilizes all the information contained in the CRS-R instead of only the highest scores on each subscale. ResultsOur findings revealed positive correlations between spectral entropy and the CRS-R index, which were more pronounced during the listening conditions than the baseline. While neural tracking of the speech envelope did not correlate with the CRS-R index, it did exhibit a positive association with CRS-R diagnoses, indicating that patients with better clinical diagnoses demonstrated higher levels of neural tracking. Additionally, we identified an interaction effect between spectral entropy and neural tracking. Specifically, higher levels of neural tracking were associated with a stronger positive relationship between spectral entropy and the CRS-R index. In contrast, when neural tracking was lower, this relationship disappeared. ConclusionThis study demonstrated the potential of neural tracking and spectral entropy as complementary tools to investigate patients with DoC. Spectral entropy proved valuable for assessing behavioral responsiveness, while neural tracking shows promise in assessing the DoC diagnosis. Terms: disorders of consciousness (DoC), neural tracking, speech envelope, spectral entropy