Paralysis can strip individuals of their ability to communicate, leaving them isolated and frustrated. This is particularly true for those who suffer from conditions like amyotrophic lateral sclerosis (ALS) or have experienced strokes affecting their brain stem.
These conditions compromise their ability to speak clearly, though they retain the knowledge of how to form words. In a groundbreaking development, researchers have harnessed brain implants to enable communication for these individuals, opening doors to more fluid interaction and expression. Two separate teams from Stanford University and UC San Francisco have documented their pioneering work in recent papers published in the journal Nature.
Unlocking Communication at Impressive Speeds
Brain-computer interfaces (BCIs) serve as the connectors between neural signals and external devices. This tech has empowered paralysed individuals to control robotic limbs, play games, and even send emails using their minds. However, translating intended speech into text or audible speech has been a challenge due to limited speed, accuracy, and vocabulary.
In the Stanford study, scientists designed a BCI employing the Utah array, a small sensor equipped with 64 electrodes. These electrodes capture neural activity, which is then decoded by an artificial neural network and translated into words displayed on a screen. The technology was tested on Pat Bennett, a 68-year-old ALS patient. The array, surgically inserted into Bennett’s cerebral cortex, successfully interpreted her attempted speech-related neural signals, enabling her to communicate at an average rate of 62 words per minute. This marks a significant leap from the previous record of 18 words per minute.
Steady Progress Towards More Communication
The UCSF team took a slightly different approach, using a surface array with 253 electrodes placed on the brain’s surface. This array detected the activity of multiple neurons across the speech cortex. Their deep-learning algorithm translated neural data collected as the patient, referred to as Ann, moved her lips. The software learned to recognise phonemes, the smallest units of language, and achieved an impressive rate of 78 words per minute—far surpassing her previous rate of 14 words per minute using a traditional communication device.
The UCSF researchers went a step further by creating a “digital avatar” to audibly relay Ann’s intended speech. By matching the avatar’s appearance and voice to Ann’s own characteristics, the team aimed to provide a more personalised and expressive means of communication for paralysed individuals.
Challenges and the Path Forward
Both approaches have their trade-offs. Implanted electrodes yield detailed information but are less stable due to shifts in the brain. On the other hand, surface arrays capture less detailed signals but offer stability over a larger area. A common challenge for BCIs is the number of electrodes that can be safely placed in the brain, impacting the clarity of the captured signals.
The road ahead involves reducing error rates and improving the longevity and reliability of implantable devices. Current error rates are still relatively high for everyday use, and BCIs must be able to record signals continuously over years without frequent recalibration. Moreover, the move towards wireless BCIs is crucial to enable seamless integration into patients’ lives.
Anticipating a Communication Revolution
While these BCIs are not yet as fast or accurate as natural speech, they represent a significant leap forward compared to existing communication systems that require manual input.
As companies like Neuralink, Synchron, and Paradromics work on wireless BCI solutions, the prospect of creating a practical medical device for patients draws closer. These breakthroughs offer hope for individuals with paralysis, promising more fluid interactions and the restoration of their voice. With continued advancements, BCIs could transform the lives of those previously silenced by paralysis.
