Nia Therapeutics has published in Brain Stimulation the first in vivo validation of its Smart Neurostimulation System (SNS), a wireless, implantable brain-computer interface designed for closed-loop treatment of memory disorders.
The SNS records neural activity from 60 channels across four brain regions—far exceeding current commercial devices. By comparison, the FDA-cleared NeuroPace RNS system for epilepsy records up to six channels, while Medtronic’s Percept adaptive DBS system records up to four.
This expanded sensing capacity addresses the distributed nature of memory, which relies on coordinated activity across widespread neural networks rather than a single focal site.
"Most brain implants were developed for conditions in which a localized abnormal signal drives symptoms," said Michael J. Kahana, co-founder and CEO of Nia Therapeutics and a professor at the University of Pennsylvania. "Decades of research show that memory depends on coordinated activity across distributed networks. The SNS was engineered to detect these patterns and respond with personalized stimulation."
In a chronic study with three sheep, the SNS demonstrated stable performance across multiple key functions -- neural-state decoding: Machine-learning classifiers accurately distinguished movement from stillness (AUC 0.92–0.98), with performance stable throughout implantation; programmable neuromodulation: Varying stimulation parameters produced predictable, dose-dependent changes in alpha- and gamma-band neural activity.
The findings show that SNS can chronically record distributed brain activity, decode behaviorally relevant states, and deliver targeted stimulation—critical steps toward chronic, closed-loop neurostimulation therapies.
The SNS builds on a decade of federally funded research, supported by DARPA and NIH. Kahana and colleagues previously recorded intracranial brain activity from hundreds of epilepsy patients performing memory tasks. Machine-learning models accurately predicted, moment by moment, whether newly learned information would be remembered.
In sham-controlled experiments, brief bursts of stimulation during classifier-identified poor-encoding states improved delayed recall by roughly 20%; stimulation at random times had no effect. Those studies established the therapeutic principle behind Nia’s approach but relied on devices unsuitable for chronic use.
"This publication shows that the core capabilities required for memory-guided stimulation—high-density sensing, real-time decoding, and programmable neuromodulation—can be delivered in a fully implantable, wireless system," said Daniel S. Rizzuto, co-founder and President of Nia Therapeutics.
