A 68-year-old woman with a debilitating neurological disorder has demonstrated the ability to communicate in real time using a brain-computer interface (BCI) implanted by the company Paradromics, marking a significant advance in the field. The wireless system, called Connexus, contains more than 400 electrodes implanted in the brain and decodes neural signals into speech using artificial intelligence (AI). The technology allows the woman to produce computer-generated spoken words by either attempting to speak or simply imagining the words without moving her speaking muscles.
The patient, whose identity has not been disclosed, suffers from primary lateral sclerosis (PLS), a condition that progressively weakens neurons that control movement related to speech, swallowing, and walking. Although she retains some ability to activate her speaking muscles, her speech is severely impaired and largely unintelligible. After receiving the implant in June at University of Michigan Health, she was able to participate in spontaneous conversations, including by phone with her grandchildren. The implanted system decodes brain activity generated by different communication-related activities such as speaking attempts, imagined speech, and listening, distinguishing distinct neural patterns associated with each.
Dr. William Marks, Paradromics’ chief clinical officer, reported that the ongoing clinical trial aims to enroll 10 patients nationwide with various disorders causing paralysis of speech and limbs, including stroke and amyotrophic lateral sclerosis (ALS). The trial operates under approval from the U.S. Food and Drug Administration’s investigational device program. Company officials anticipate further clinical trials with larger patient groups before seeking full FDA approval for commercial use.
Paradromics founder and CEO Matt Angile emphasized the system’s ability to interpret complex neural signals related to different forms of communication. Vikash Gilja, the company’s chief scientific officer, noted the system’s performance remained effective when the patient whispered—the less physically demanding mode—or used imagined speech, although decoding clarity improved with practice. The capability to differentiate signals generated from silent internal thoughts, attempted speech, and auditory inputs is key to ensuring patients maintain control over when they communicate.
Experts not involved in the study praised the results as a meaningful step toward practical clinical tools for restoring speech in people with severe paralysis. Dr. Edward Chang, chairman of neurological surgery at the University of California, San Francisco, highlighted the importance of the wireless and self-contained design. Previous approaches often required patients to be physically tethered to a computer via a cable, which presented infection risks and limited mobility.
Dr. Chang’s own research, published recently alongside the Paradromics announcement, demonstrated the potential to decode signals simultaneously for both speech and physical gestures using implanted electrodes, potentially enabling patients to control comprehensive digital communication interfaces. Paradromics and similar companies use surgically implanted electrode arrays beneath the brain surface to maximize signal clarity, whereas other efforts focus on less invasive methods such as catheter-delivered electrodes within blood vessels. Each approach balances considerations of surgical risk, long-term stability, affordability, and signal specificity.
While some implanted electrode arrays have encountered challenges related to tissue damage and signal degradation over time, sustained long-term performance remains a critical milestone on the path to clinical adoption. Paradromics’ trial is ongoing, with continued evaluation of the system’s durability and effectiveness in improving communication capabilities for individuals with paralysis.