About 20% of US adults live with chronic pain, an estimated 50 million people, according to CDC figures cited in a New York Times report on personalized brain stimulation. For the small group whose pain survives every drug, nerve block, and spinal implant, a brain implant sounds like the last door. That door is still being built.
This technology is medical, not consumer, and it sits at the early clinical trial rung. The strongest recent result comes from a feasibility study of six people. Here are five things worth knowing before the idea turns into a hope.
1. No brain implant is approved for chronic pain
Deep brain stimulation (DBS) has had FDA approval for Parkinson’s disease since 2002. Pain is a different story. A 2025 review in Frontiers in Medicine says DBS for pain has stayed off-label since the FDA approval was withdrawn after the late 1970s. A 2023 systematic review adds that the push for approval ended after two multicenter studies. UCSF’s Prasad Shirvalkar told the New York Times that results were inconsistent, with some patients getting no help and others losing relief over time.
Here is where each approach sits. Conventional DBS for pain is an off-label procedure at specialist centers. Closed-loop DBS for pain is an early feasibility study run under an FDA Investigational Device Exemption. Nothing in this category is a product you can buy. Policy outside the US is no looser: as of its 2015 clinical commissioning policy, NHS England did not routinely commission DBS for chronic neuropathic pain.
2. The best recent result is six people and a preprint
The groundwork was a 2023 Nature Neuroscience study from Shirvalkar and Edward Chang’s group at UCSF. They implanted electrodes in the anterior cingulate and orbitofrontal cortex of four people and predicted chronic pain severity from orbitofrontal activity. The follow-up is the PRISM trial, posted as a medRxiv preprint in August 2025. It had not been through peer review when posted, and I could not find a peer-reviewed version. The design:
Participants: six adults aged 48 to 58 with refractory pain (three post-stroke, one chemotherapy-induced neuropathy, one complex regional pain syndrome, one spinal cord injury).
Mapping phase: 10 days of temporary electrodes, with up to 14 candidate sites per hemisphere tested for stimulation targets and pain biomarkers.
Permanent implant: five participants received two Medtronic Summit RC+S systems and four 4-contact leads each, sensing at 250 Hz on four bipolar channels per device.
Follow-up: up to 3.5 years.
In the double-blind crossover phase, closed-loop stimulation cut average daily pain by about 50% (range 12 to 87%). Sham was associated with a mean 11% increase. Step counts rose 18% versus 1%, and the system stimulated only 7 to 55% of the time, where conventional DBS is always on. News coverage cited roughly 60%, but the preprint’s group figure is the lower number, so use that one.
The caveats decide how much weight this carries. Sham testing happened in four of five participants because one declined it. Nobody compared closed-loop with continuous stimulation, so the study does not show that the “thermostat” matters, a point Shirvalkar and Cleveland Clinic’s Andre Machado both made to the Times. The authors also hold a patent on closed-loop DBS for pain, and Medtronic donated the hardware without a role in the analysis, per the preprint. For the full research, funding, and trial picture behind this story, NeurotechMag Pro has you covered.
If you are building in this space, the open question is the one the authors flag: is biomarker-driven control necessary, or would timed stimulation do just as well?
3. Older brain targets have a thin controlled record
Before PRISM, the main targets were the sensory thalamus, periaqueductal grey, anterior cingulate, and motor cortex. A critical review of motor cortex stimulation covered 14 studies and 210 patients, all case series and none controlled. About 55% of operated patients had a good response. A later Stanford series of 14 patients found only two still above 50% improvement at their last visit.
Cleveland Clinic ran one of the few randomized trials, in ten patients with post-stroke pain. The blinded phase showed no significant difference on the main disability measure, though affective measures improved, and the team planned a multicenter follow-up. A meta-analysis of 43 studies and 966 patients found a significant effect in people who received DBS for pain. The PRISM authors note that most earlier work was unblinded or lacked a placebo arm.
My read is that the field has more than 65 years of impressive case series and disappointing trials. PRISM is better designed than most of them, and it is still six people.
4. The risks come from the surgery, and they are not small
NICE’s procedure guidance cites a case series of 141 patients with infection in 12%. Another series of 122 patients reported ventriculitis in one, subgaleal infection in four, and subdural empyema in one, and two of those patients needed the system removed. NICE’s specialist advisers also listed mood change from aberrant stimulation, speech difficulty, and lead migration. The motor cortex review above reported early postoperative seizures in 12% of patients.
PRISM reported serious surgical adverse events in two of six participants. One had an infection that required removing the device on one side of the brain, and another had speech problems that took months to resolve. A third participant had a brief seizure during the exploratory phase. The investigators reported no serious events from the stimulation itself.
5. Know what else is on the table, and what to ask a trial team
A brain implant is the last step, not the first. Spinal cord stimulation (SCS) is the nearer implant, and its evidence is contested. A Cochrane review of 13 trials with 699 participants found SCS probably does not improve low back pain against placebo at six months. It does not support using SCS for back pain outside clinical trials. A 2026 network meta-analysis of 16 randomized trials found low to moderate certainty that SCS beats conventional medical management, and rated all but one trial at high risk of bias. The safest reading is that SCS looks better than usual care and is unproven against sham.
A non-invasive option is at pilot stage. A low-intensity focused ultrasound trial for generalized chronic pain listed an estimated primary completion of March 31, 2026, and I found no results yet.
On access, UCSF’s deep brain stimulation trials page listed a closed-loop pain study as open to eligible people aged 22 to 80 when I checked. PRISM enrolled people who had failed every therapy, rated baseline pain at 6 out of 10 or higher, and had no untreated psychological disorders. If you or someone you know is weighing a trial, a pain specialist should map the options first. Questions to bring to any team:
How long is the sham phase, and when does a participant learn which condition they received?
What happens to the implanted hardware after the study ends?
What are the rules for removal if there is an infection or the device does not help?
What psychiatric and rehabilitation support comes with the surgery?
So here is the question for the field. Should a six-person preprint with a 50% average reduction be enough to justify a multicenter trial, or does it need a continuous-stimulation arm first?