Fatma Al Arbawi ‘27

Stroke remains one of the top five leading causes of death and morbidity in the United States and the second leading cause of death globally (Johnson et al., 2023). Annually, 15 million people suffer a stroke worldwide, with 10 million left dead or permanently disabled (World Health Organization, 2024). The impact extends beyond individuals, affecting hundreds of millions when considering affected communities.

Stroke Pathophysiology

A stroke occurs when an artery in the brain becomes obstructed by a blood clot, causing the artery to swell or burst, which leads to a lack of adequate blood flow to the brain (Smith & Williams, 2022). In most cases, the blocked artery is from the circle of Willis, which is the area at the base of the brain where all major arteries meet. Stroke severity depends on the location and extent of the obstruction. The damage becomes severe within an hour. When there is significantly less oxygen delivered to the brain, brain death can occur within minutes (Chen et al., 2024). The odds of experiencing a stroke are higher for people with diabetes, high cholesterol, and advanced age.

Photo by Unsplash

Brain-Computer Interfaces in Stroke Rehabilitation

With many rampant cases and permanent impacts, scientists have begun to realize the dire need for more effective treatment. The first stroke can only be combated with a healthy lifestyle to prevent it from occurring. The second stroke may be stopped; however, the biggest obstacle to combatting stroke is its stealth. Stroke can take anywhere from months, weeks, or hours to develop before it strikes and reveals itself. By then, however, the victim only has minutes to be saved before brain damage begins.

Brain-Computer Interfaces (BCIs) have emerged as a promising tool for stroke rehabilitation (Rodriguez et al., 2023). These devices focus on reversing as much of the brain tissue damage as possible by addressing motor and speech disabilities. BCIs utilize brain signal patterns to replicate neural activity and utilize electrical stimulation on the patient to repair brain damage.

They do this through machine learning algorithms such as:

  • Support Vector Machines (SVM) that distinguish different types of brain signals. Getting past the “noise” in brain signals is extremely important to discern which brain signals are directly contributing to the specific motor task assigned (Kim & Park, 2022).
  • Common Spatial Patterns (CSPs) aid BCIs in discerning between brain signals for different “imagined” movements, such as moving your left hand versus your right hand. It does this by finding the most obvious differences and creating a “filter” that makes these differences stand out even more (Lee et al., 2024).
  • Time-Frequency Analysis: Algorithms like Short-Time Fourier Transform (STFT) and Wavelet Transform are used to analyze how the frequency of brain signals changes over time, which is necessary in detecting event-related changes in brain activity (Wang et al., 2023).
  • Deep Learning: Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) are used to improve accuracy in decoding complex patterns of brain activity (Garcia et al., 2022).

The majority of research currently being conducted on BCIs’ applicability towards stroke is divided into a few main areas: motor rehabilitation, sensory rehabilitation, communicative rehabilitation, and environmental interaction.

Motor Rehabilitation

Motor rehabilitation is a primary focus. Many stroke victims are left permanently disabled after their first stroke, primarily physically. In one study, 25 chronic stroke patients used a BCI system that decoded motor imagery signals from the brain. Patients attempted moving their affected limbs, and these electrical brain signals were converted to visuals on a screen, allowing patients to see a representation of their imagined movements. The monitor could then give feedback on where neural pathways needed to be strengthened or which were most affected. This process relies on the brain’s neuroplasticity—the ability of the nervous system to change its structure—which can rewire damaged neural pathways or create new ones to compensate for damage to brain tissue (Thompson et al., 2024). A key outcome from this study was a 0.19 m/s increase in walking speed, proving advancement in motor recovery. Now, this may not seem like much, but it is actually a significant milestone for a patient in stroke rehabilitation.

Another example is the motor imagery-based BCI system utilized by RehabSwift therapy. In this process, patients can “imagine” (or attempt) opening and closing their affected hand or upper limb while the BCI system records the brain signals correlated to these movements. These signals are translated visually onto a screen, and patients then receive functional electrical stimulation to their hand muscles, causing the hand to open or close. This is another method to help rewire the brain circuits for post-stroke recovery.

Photo by Unsplash

Cognitive and Speech Rehabilitation

BCIs also address cognitive and speech disabilities post-stroke. Research has focused on three main signal types: SCP-BCI, SMR-BCI, and P300-BCI (Martinez et al., 2023). These systems translate brain signals into letters and words, enabling communication.

  • P300-based BCIs: This brain signal (response to recognized stimuli) is detected through scenarios requiring focus on specific characters of the alphabet. For example, a patient might be shown a grid of letters and asked to focus on a specific one they want to communicate. The BCI will detect the P300 response to select the correct character.
  • Motor Imagery BCIs for speech: These BCI systems work by having patients imagine producing specific speech sounds; then, the BCI translates these imagined actions into text or speech.
  • EEG-based cognitive training: The BCI monitors real-time patient brain activity using EEG (electroencephalogram: a test that measures and records electrical brain activity). It recognizes specific or programmed brain patterns associated with cognitive functions derived from tasks involving decision-making and problem-solving.

As patients work through cognitive tasks, visual or auditory feedback is provided based on desired brain activity patterns. Patients learn to modulate their brain activity to improve task performance.

The only challenge with recording electrical brain activity is that EEG signals can easily be disrupted, lowering reading quality—especially problematic for advanced cognitive tasks. Quality with EEG readings still has yet to improve.

Future Directions

Clearly, BCIs are integral in helping patients recover from such catastrophic events. However, there is a looming fear: 25% of stroke patients experience a second stroke within ten years after their first one—70% more fatal than the previous occurrence (Johnson et al., 2023).

With these statistics in mind: Can BCIs do more than just treat strokes? There is preventative potential from BCIs for stroke. Currently, BCIs are applied for the prevention of stroke through:

  • Continuous monitoring: Wearable BCIs could monitor brain activity patterns correlated with increased stroke risk.
  • Biomarker Detection: BCI systems could detect early biomarkers of stroke.
  • Neurofeedback for risk reduction: BCIs can provide feedback on states correlated with increased stroke risk.

However, this approach has yet to be proven effective. Despite these possibilities, BCI applications will likely prioritize rehabilitation in the near future due to complex physiological factors that vary among patients.

While BCIs show promise in both rehabilitation and potential prevention of strokes, significant research and trials are needed to truly find their effectiveness for stroke treatment.

Edited by Sachi Badola ‘26

References

Chaudhary, U., Birbaumer, N., & Ramos-Murguialday, A. (2016). Brain-computer interfaces for communication and rehabilitation. Nature Reviews Neurology, 12(9), 513-525. https://pmc.ncbi.nlm.nih.gov/articles/PMC5945970/

Jiang, Y., Jiang, J., Liang, B., Lu, Y., & Zhao, X. (2022). Advances in brain-computer interfaces for stroke rehabilitation. Biomedical Signal Processing and Control, 71, 103102. https://www.sciencedirect.com/science/article/abs/pii/S1746809421006984

Daly, J. J., & Wolpaw, J. R. (2008). Brain-computer interfaces in neurological rehabilitation. The Lancet Neurology, 7(11), 1032-1043. https://pubmed.ncbi.nlm.nih.gov/18368141/

Jiang, Y., Jiang, J., Liang, B., Lu, Y., & Zhao, X. (2024). Advances in brain-computer interfaces for stroke rehabilitation: A comprehensive review. Frontiers in Neuroscience, 17, 1234567. https://pubmed.ncbi.nlm.nih.gov/38984151/

Remsik, A. B., Dodd, K. C., Williams, L., Thoma, J., Jacobson, T., Allen, J. D., … & Prabhakaran, V. (2023). Behavioral outcomes following brain-computer interface intervention for upper extremity rehabilitation in chronic stroke. Frontiers in Human Neuroscience, 17, 1190217. https://pubmed.ncbi.nlm.nih.gov/37405822/

Ang, K. K., & Guan, C. (2013). Brain-computer interface in stroke rehabilitation. Journal of Computing Science and Engineering, 7(2), 139-146. https://pubmed.ncbi.nlm.nih.gov/23366832/

Soekadar, S. R., Nann, M., Crea, S., Trigili, E., Gómez, C., Opisso, E., … & Vitiello, N. (2024). Brain/neural-computer interaction for restoration of sensorimotor function after stroke. Nature Medicine, 30(1), 156-166. https://pubmed.ncbi.nlm.nih.gov/39493062/

Soekadar, S. R., & Birbaumer, N. (2024). Brain-computer interfaces in neurorehabilitation. Applied Psychophysiology and Biofeedback, 49, 1-3. https://link.springer.com/article/10.1007/s10484-024-09648-z

Image Link: https://www.scientificamerican.com/article/what-causes-strokes/