Investigating how cognitive and motor functions interact in people with multiple sclerosis (PwMS) is key to identifying disease-related alterations, yet technical challenges have long hindered this research. We introduce an innovative framework for dual-task (DT) assessment in PwMS, integrating wearable electroencephalography (EEG) and inertial measurement unit (IMU) sensors. The study involved 11 PwMS (6 males, 5 females; mean age: 58.5 ± 12.5 years), who performed the timed up and go (TUG) test under two conditions: a motor task as single-task (ST) and a motor-cognitive DT. A custom-made bracelet including an IMU and a wearable EEG device allowed the recording of motion data and brain activity during the tasks. The former was used for automatic trial segmentation through a hybrid convolutional neural network and a long-shortterm-memory model. This allowed the extraction of the EEG epoch during the task. Finally, the power spectral density in θ (4-8 Hz) and α (8-12 Hz) frequency bands was analyzed to extract mental workload. The proposed segmentation method achieved accurate estimations when compared to manual video-based labeling, with a very strong correlation R2=0.993 and mean-absolute-error of 0.383 s and 0.294 s for the ST and DT conditions, respectively. The EEG results showed increased mental workload during the DT condition (Wilcoxon signed-rank test p-value = 0.02). The planned inclusion of new participants will help to confirm the robustness of cortical activity patterns during DT performances in PwMS.Clinical relevance-The proposed experimental protocol represents a significant step forward to the understanding of neural correlates of cognitive-motor interaction, allowing the recording of cortical activity during walking. These findings could contribute to a better understanding of the association between DT performance and risk of falls in PwMS and help design specific and personalized rehabilitation interventions.
