US firstly approves human trials for speech-restoring brain implant

An American neurotechnology company has received regulatory approval for the first clinical trials of Connexus, a fully implantable brain–computer interface (BCI) designed to restore speech in patients with severe paralysis.

It is the first device of its kind in the world to receive authorization for human testing, according to its manufacturer, Paradromics, based in the US state of Texas. The system is entirely wireless and could mark a major breakthrough for patients who have lost the ability to speak, as reported by New Atlas.

In June of this year, Paradromics tested its brain implant on a living human for the first time. In that initial experiment, the device was implanted in a brain region scheduled for surgical removal. The goal was to evaluate material compatibility and sensor performance in human tissue. The test was deemed successful, enabling the company to apply for clinical-trial approval.

The US Food and Drug Administration (FDA) has now authorized Paradromics to implant Connexus sensors in two patients, one device per participant. If the trial proves successful, the study size will expand to 10 people, with two of them receiving dual implants to increase data throughput and improve the accuracy of neural monitoring.

Each Connexus module is roughly the size of a small coin and equipped with over 400 platinum–iridium electrodes. Each electrode is about 40 microns thick, only slightly larger than a single neuron. This design allows the device to capture activity from individual neurons in the motor cortex — the core advantage of the Connexus platform.

Paradromics emphasizes that the device cannot read a person’s thoughts. Instead, it records neural activity produced when a person mentally attempts to speak — generating motor commands for the larynx, tongue and lips. For people with paralysis, no actual muscle movement occurs and the person remains unable to speak, but the brain still produces the neural signals associated with attempted speech. The device captures this activity and, with the help of a personalized large language model, converts it into spoken output. If archival audio recordings of the patient are available, the synthesized voice can closely mimic the patient’s natural speech.

The electrodes will be inserted a short distance into the motor cortex. The controller that processes neural activity is integrated directly into the implant. The processed signal then travels through a thin subcutaneous cable to the patient’s chest, where a small optical transceiver is implanted. This transceiver wirelessly sends the data to an external wearable receiver. That receiver also acts as a wireless charging unit and connects to a small computer running decoding software and a speech synthesizer.

According to the publication, the study will also evaluate whether the system can decode imagined hand movements, which could make it possible for patients to control a computer cursor or other digital devices. These first human clinical trials bring the field closer to a future in which patients who have lost speech due to stroke or injury may regain full communication abilities.

By Nazrin Sadigova