June 05, 2025
Over the last 11 years, research focused on the use of neuroprostheses to restore lost function in individuals with spinal cord injuries (SCIs) has made tremendous strides forward. Broadly speaking, neuroprostheses are devices that use electricity to directly communicate with the nervous system or targets innervated by the nervous system.
In this Q&A, Ryan J. Solinsky, M.D., describes the rapidly evolving use of neuroprostheses and their role in treating individuals with SCIs. Dr. Solinsky is a physiatrist and scientist at Mayo Clinic in Rochester, Minnesota, whose research is aimed at deciphering autonomic neurophysiology and leveraging neurotechnology to enhance the quality of life for individuals with SCIs.
How is spinal cord stimulation changing what we define as possible for SCI recovery?
Implanted epidural spinal cord stimulation
In a 2018 publication in Nature Medicine, Mayo Clinic researchers and others demonstrated that implanted epidural spinal cord stimulation allowed individuals with spinal cord injuries to walk with some assistance.
In a 2018 publication in Nature Medicine, Mayo Clinic researchers and others demonstrated that implanted epidural spinal cord stimulation allowed individuals with spinal cord injuries to walk with some assistance. Spinal cord stimulation is also being used to help improve patients' sitting, reaching and trunk stability.
More recently, researchers have focused on using noninvasive spinal cord stimulation to improve hand function. In a 2024 Nature Medicine publication, researchers showed that patients who received a combination of occupational therapy and transcutaneous spinal cord stimulation experienced improvements in upper extremity muscle strength and grasp force. Collectively, these rewrite what we would otherwise have told patients about expected recovery even 10 years ago.
Is spinal cord stimulation being used for nonmotor targets?
Motor and nonmotor targets behave very differently. A few research groups are exploring whether closed-loop stimulation can be used to target nonmotor outputs, including sensory control and blood pressure control. One of our goals is to determine whether nonmotor targets controlled by the autonomic nervous system respond to stimulation and result in improved autonomic regulation or whether this intervention simply causes low-grade autonomic dysreflexia. The data available to date suggests that while the spinal cord stimulation can statically activate the sympathetic nervous system, it's not necessarily improving control of these nonmotor targets.
How are researchers working to deliver more targeted spinal cord stimulation?
Researchers are moving from a static system like the transcutaneous spinal cord stimulation that is just on and just blasts the whole time, to delivering the right stimulation at the right time. Some efforts are focusing on the use of brain-computer interfaces that capture an individual's intentions, the input signals from the brain, which are then used to pattern and deliver the stimulation in ways that optimize motor control for whatever the task is. Methods being studied involve the use of an implanted electrode array that can record direct neural signals from the surface of the brain or inserted 1 to 2 millimeters deep. Another approach involves the use of an endovascularly deployed stent containing multiple electrode leads, eliminating the need for brain surgery.
Can you describe how machine learning and artificial intelligence might help make the delivery of spinal cord stimulation more targeted?
SCI researchers are tapping into the potential of closed-loop machine learning algorithms to optimize the effectiveness of spinal cord stimulation. Closed-loop machine learning algorithms are trained on large data sets of recordings and then use that information to determine the optimal output stimulation to produce the desired result. Over time, the model's outputs can be measured, creating a constant feedback loop the allows the model to adjust and refine its recommended output.
This process is challenging in patients with spinal cord injury because the inputs and output targets are dynamic. As a result, the field is shifting toward the use of dynamic architectural neural networks that can accommodate and change the stimulation.
What are some of the more common neuroprosthesis design pitfalls and challenges, and how can insight from PM&R clinicians help overcome these?
The design focus of neuroprostheses sometimes fails to address the complex pathophysiologies that our patients have. PM&R clinicians have unique skill sets that allow us to see the entire patient. We bring a richer understanding of what the pathophysiology really means, and we understand that our patients are interested in more than how fast they can walk 10 meters. We talk to people about issues such as bowel control, spasticity, neuropathic pain and how that can impact their quality of life. These issues and tasks such as transfers, having enough grip to feed themselves and to perform more-independent self-care are goals that we can give to engineering teams to help them identify the targets with the biggest impact.
We also can help provide patients with better expectations of what they might experience with these interventions. Although there are many positive results associated with spinal cord stimulation, not every individual is going to experience all of those benefits. As the field becomes more mature and we start to understand those differences and how to predict them, we can help share that information with patients.
For more information
Gill ML, et al. Neuromodulation of lumbosacral spinal networks enables independent stepping after complete paraplegia. Nature Medicine. 2018;24:1677.
Mortiz C, et al. Non-invasive spinal cord electrical stimulation for arm and hand function in chronic tetraplegia: A safety and efficacy trial. Nature Medicine. 2024;30:1276.
Refer a patient to Mayo Clinic.