Personalized upper limb stroke rehabilitation using data-driven multi-modal electroencephalography (EEG) and near-infrared spectroscopy (NIRS) based brain computer interface (BCI) with soft robotic glove

Abstract
Background
Motor impairments, particularly in the upper limbs (UL), significantly diminish the quality of life for stroke survivors by impacting their ability to perform daily activities and convention rehabilitation strategies have yet been optimized. The use of brain-computer face (BCI)-based motor imagery concomitant with physical practice presents an alternative. This study aim to develop a data-driven personalized stroke rehabilitation using the EEG-based BCI Soft robotic glove intervention with online NIRS monitoring.
Methodology
An EEG and NIRS based BCI uses motor imagery (MI) by a stroke participant to trigger a soft robotic glove to carry out 6 interactive activities of daily living (ADLs) intervention. 4 chronic stroke participants (Fugl-Meyer Motor Assessment-Upper Extremity, FMA-UE, score 11-27, mean age, 58.5 years, mean post-stroke durations, 4.0 years, 2 Male, 3 infarcts) were pre-screened with motion capture (MOCAP) analysis on 6 UL tasks that were matched in motion similar to the 6 BCI ADLs interventions and subsequently prescribed a personalized intervention based on their MOCAP performance. They then underwent 18 sessions x 120 minutes over 6 weeks of intensive upper limb BCI training administered by a Physiotherapist.
Results
Performance efficacy was measured using FMA-UE, Action Research Arm Test (ARAT) scores at week 0, week 6 and week 12. All 4 patients completed MI-BCI SR training without issues and demonstrated improvement both outcome measures. Mean FMA-UE scores at week 0, 6 and 12: 25.3 ± 7.7, 32.0 ± 9.3, 34.0 ± 9.8. Mean ARAT scores at week 0, 6 and 12: 15.5 ± 7.7, 20.3 ± 9.7, 20.8 ± 9.7
Conclusion
There were overall improvements in the upper limb motor performance of the stroke participants with carry over effects at 12 weeks. Hence EEG-based BCI-SR hence is potentially a viable and effective tool for upper limb stroke rehabilitation with potential for further optimization of UL motor performances in chronic stroke populations.