A feasibility study involving five people with Parkinson’s disease found that a form of deep brain stimulation adjusted in real time to match each person’s walking pattern reduced falls and improved some gait measures compared with standard continuous stimulation.
Walking difficulties are among the most limiting symptoms of Parkinson’s disease. Patients commonly develop uneven steps, poor balance, and freezing episodes that raise the risk of falls. Conventional deep brain stimulation — which delivers constant electrical pulses to brain structures — manages tremor, stiffness, and slowness reasonably well, but its effects on walking tend to be inconsistent and can erode over time. Attempts to tune stimulation through different frequencies or electrode positions have often improved gait only by worsening other motor symptoms.
That gap prompted researchers to test whether stimulation could instead respond dynamically to movement as it happens.
What Was Already Known
Implanted electrodes in deep brain structures can record electrical activity — called local field potentials — that shifts with movement, medication, and sleep. Earlier adaptive deep brain stimulation (aDBS) systems have used these signals to modify stimulation automatically, but most relied on slowly changing markers of overall disease state. Walking presents a tighter constraint: each stride lasts roughly one second, so a system tracking gait must detect movement and respond almost instantly.
How the Study Was Conducted
Researchers at a single center enrolled five participants with Parkinson’s disease. Each received electrodes implanted in the globus pallidus internus, a structure targeted in standard DBS surgery, along with cortical electrode paddles placed beneath the skull — a step not part of routine DBS procedures. Both sets of implants connected to an investigational bidirectional device called the Medtronic Summit RC+S, which can stimulate the brain and record neural activity simultaneously.
Before programming the adaptive system, participants walked at their preferred pace for at least 250 steps. Researchers analyzed those recordings to identify which brain signals corresponded to specific phases of each person’s stride. They then programmed the device to shift stimulation amplitude — between half and full therapeutic level — during the swing phase of the opposite leg, using one neural feature per hemisphere selected for each individual.
Three of the five participants went on to complete a randomized, double-blind crossover trial lasting several days in their normal daily environments. They cycled through three conditions: continuous stimulation; adaptive stimulation that increased from half to full amplitude during the swing phase; and adaptive stimulation that decreased from full to half amplitude during the same phase. Neither participants nor evaluating clinicians knew which mode was active at any given time.
What the Results Showed
The researchers successfully identified individualized gait-related brain signals in all five participants and integrated them into the adaptive system. In short-term clinical testing, the adaptive approach improved step variability and step symmetry compared with conventional continuous stimulation.
During the multi-day blinded crossover phase, participants experienced fewer falls while using adaptive stimulation. Overall control of Parkinson’s motor symptoms remained stable across all conditions. Participants tolerated the rapid stimulation changes without serious adverse events.
The extent of gait improvement varied from person to person, consistent with the individualized design of the system.
What the Researchers Concluded
The team characterized the work as a demonstration of technical feasibility rather than a proof of clinical benefit. Senior author Doris D. Wang, MD, PhD, noted that walking requires highly coordinated timing across both sides of the body, and that the new system recognizes movement patterns and adjusts stimulation in real time to work alongside natural movement. First author Kenneth H. Louie, PhD, observed that the brain contains rich movement-related information, and that identifying neural signatures tied to each step made it possible to guide stimulation dynamically.
The investigators called for larger randomized clinical trials to determine whether the approach produces meaningful long-term benefits.
Several factors limit what can be concluded from this study. The sample consists of five participants from a single center; only three completed the multi-day blinded phase. That number is too small to establish clinical effectiveness. The study was designed explicitly as a feasibility trial, not an efficacy trial.
The neurostimulator used is investigational and not commercially available. The cortical electrode paddles are not part of standard DBS surgery, adding procedural complexity not present in typical implantation. Gait responses differed across participants, which restricts generalization to the broader Parkinson’s population. Long-term follow-up data are not yet available. Some authors also disclosed consulting relationships and employment affiliations with neuromodulation companies.
Reference
Kenneth H. Louie, Jannine P. Balakid, Jessica E. Bath, Seongmi Song, Hamid Fekri Azgomi, Jacob H. Marks, Julia T. Choi, Philip A. Starr, Doris D. Wang. “Adaptive deep brain stimulation for dynamic gait control in Parkinson’s disease: a randomized feasibility trial.” Nature Medicine, June 15, 2026. DOI: 10.1038/s41591-026-04434-2
