Artificial Touch, Measured Over Ten Years

The hard question for a sensory neuroprosthetic was never whether stimulation produces a sensation. Passing current through the somatosensory cortex and evoking a feeling in a hand the person can no longer feel has been demonstrated for years. The question is whether it still works in year eight, with the same electrodes, at a current the tissue tolerates. Almost every published result has been a snapshot, and snapshots cannot answer a question about decay.

A group across the University of Pittsburgh, the University of Chicago and Northwestern has now reported the long view. Five people with spinal cord injury, each implanted with two microelectrode arrays in the hand region of Brodmann's area 1, were followed across implant durations of two to ten years. The combined total is 27 patient-years and more than 168 million stimulation pulses, delivered with no serious adverse events and no measurable degradation of electrode health.

Why it matters: Every sensory prosthesis, for touch or hearing or vision, rests on an unproven assumption that implanted electrodes keep doing their job for the decades a patient will live with them. This is the first dataset large enough in time to test that assumption rather than assert it.

What the trial actually did

Each participant received two Blackrock NeuroPort arrays, grids of silicon electrodes a few millimetres across, placed in the hand representation of Brodmann's area 1. That is the strip of somatosensory cortex where touch from the skin arrives, laid out as a map of the body, which is what makes the approach possible at all: stimulate the part of the map corresponding to a fingertip and the person feels the fingertip. The measurements taken over the following years were deliberately mundane. Detection thresholds, the projected field of each electrode, participant reports of what the sensation felt like, and the electrical health of the array itself.

What actually degrades, and how fast

Two things could fail over a decade, and they fail differently. The hardware can deteriorate, in which case electrode impedance and signal quality drift and the array becomes unusable. Or the tissue can respond, walling off the electrodes in glial scar so that progressively more current is needed to reach neurons that are still perfectly healthy. The second is the more insidious problem, because it looks like a working device that quietly demands more of the brain each year.

The measured drift was small. Detection thresholds, the minimum current needed for a participant to notice a sensation, rose by roughly 3.5 microamperes per year. Against stimulation currents in the tens of microamperes that is a slow slope, and it is a number rather than a reassurance, which is the point of running the study this long. Functional electrodes declined by about 21 percent, leaving 64 percent still reliably evoking touch. In the participant at the ten-year mark, 60 percent were still working.

The sensations stayed put

Durability of hardware would be worth little if the percepts themselves wandered. A prosthetic hand that reports pressure to a location that drifts from fingertip to palm over five years is not a usable device, and cortical reorganization after injury made drift a reasonable thing to expect. It did not happen. As Greenspon and colleagues report in Science Translational Medicine, both the quality of the evoked sensations and their projected fields, the specific patches of hand where each electrode is felt, remained consistent. Their summary is that the approach was safe and effective, consistently evoking informative somatosensory percepts as long as 10 years.

The adverse events that did occur were minor and worth stating precisely: sensations occasionally persisted briefly after stimulation stopped, between 3 and 25 such events per participant across years of use. That is a rate low enough to characterize rather than a safety signal.

What the study can't say yet

Five participants is a feasibility cohort, not a population. Threshold drift of 3.5 microamperes per year is an average across electrodes and people, and the variance matters more than the mean when the question is whether a given patient's device will last; a single participant carried the ten-year observation. Nothing here establishes that the twentieth year resembles the tenth, and the failure mode for implanted electrodes has historically been that degradation is not linear.

The study also measures the delivery of sensation, not its usefulness. Evoking a reliable percept localized to a fingertip is a precondition for a prosthetic hand that feels like a hand, and it is not the same as demonstrating that people manipulate objects better because of it. That is a separate claim requiring separate evidence, and this paper does not make it.

Nor is the cohort representative in an ordinary sense. These are participants in a long-running early feasibility trial, closely followed at specialist centres, with arrays placed by experienced surgeons. Device longevity measured under those conditions is an upper bound on what a wider clinical rollout would see, not an estimate of it.

Quick questions

Does this mean brain implants are safe long-term? It means these electrodes, in this cortical location, in five people, produced no serious adverse events over up to a decade. That is meaningful evidence for this application and does not generalize to implants elsewhere in the brain.

Why does the threshold rise at all? The likeliest explanation is the tissue response around the electrode rather than hardware failure, since electrode health measures stayed stable. The paper documents the drift without attributing a mechanism.

What's the one-line takeaway? Across 27 patient-years and 168 million pulses, cortical stimulation kept evoking touch in the right place, with detection thresholds creeping up only about 3.5 microamperes a year and 60 percent of electrodes still working after ten years.