Deep brain stimulation has been used to treat Parkinson’s disease for decades. Electrodes are implanted in a small structure deep in the brain called the subthalamic nucleus, a pulse generator sits under the collarbone, and for many patients the tremor and rigidity ease substantially. What has never been fully settled is why it works — and, more practically, why it works much better for some patients than others.

A study published in the journal Brain offers an answer with an unusually specific number attached: a rhythm between 20 and 35 hertz.

What was found

The work came from a team spanning the University Hospitals of Cologne and Düsseldorf, Harvard Medical School and Charité Berlin. They studied 50 patients — 100 brain hemispheres — recording activity through the implanted stimulation electrodes and simultaneously using magnetoencephalography, which measures the magnetic fields produced by brain activity from outside the skull.

What emerged is that the benefit of stimulation appears to depend on engaging a particular network connecting the subthalamic nucleus to frontal regions of the cortex, and that this network communicates mainly in the fast, or high, beta range. The strength of that connection tracked closely with how much each patient’s motor symptoms actually improved after implantation.

Bahne Bahners of Düsseldorf University Hospital, the study’s first author, described the rhythm as functioning like a communication channel between the deep structure and the cortex, potentially mediating the therapeutic effect. Andreas Horn of the University of Cologne, who led the study, said it was the first characterisation of the response network in these terms.

Why combining two methods mattered

The genuinely novel part is methodological, and it explains why this took so long.

Two research traditions have been circling this question in parallel. Functional MRI studies map which brain regions need to be connected to a stimulation site for patients to improve — excellent spatial precision, but far too slow to capture brain oscillations, which happen on a millisecond scale. Electrophysiology, recording from implanted electrodes, has the opposite profile: it resolves the timing beautifully but samples only a handful of locations.

Nobody had captured both the where and the when at once. Recording from the electrodes while simultaneously running magnetoencephalography does exactly that.

There is also a wrinkle worth noting for anyone who follows this field. Beta activity has long been the prime suspect in Parkinson’s, but most attention has gone to the lower part of the band, roughly 13 to 20 hertz, which has been most consistently linked to rigidity and slowness of movement. This study points at the faster end of the range instead — not overturning the earlier work, but suggesting the part of the beta band that signals disease severity and the part that mediates treatment response may not be the same part.

What it could change

Programming a deep brain stimulator is currently a slow, largely trial-and-error process. A clinician adjusts contacts, amplitude and frequency across repeated appointments, watching how the patient responds. If a specific network and a specific rhythm predict who benefits, those settings could in principle be targeted directly — which would matter most for patients whose implants have not delivered the improvement they hoped for.

The appropriate caution: this is a finding about mechanism, not a new treatment. It has not yet been shown that programming to this target improves outcomes in practice, which would require its own trial. Fifty patients is a solid sample for this kind of intensive recording work but modest for clinical conclusions. Nobody’s stimulator settings change because of this paper.

What it does offer is a plausible physiological explanation for something that has been observed for years, and a measurable target for the next round of work. In a field where a great deal of adjustment happens by feel, that is a meaningful step.

Sources

Bahners et al., The deep brain stimulation response network in Parkinson’s disease operates in the high beta band, Brain, 2026; University of Cologne announcement via ScienceDaily; associated coverage in specialist neurology press.