Suppose a stroke or trauma causes you to become paralyzed from your crown to your toes overnight. Only the muscles that control your eye movement still work. Your brain function works as it always did – you can form thoughts and reasonings but can’t express them.
The head and body turn into a prison for patients who lose the ability to communicate while conscious. To free people from this isolation, scientists and entrepreneurs – with Elon Musk’s Neuralink being the biggest – are working on technology to translate brain signals into computer commands. And they are succeeding more and more.
- Scientists and entrepreneurs are developing technology to translate brain signals into computer commands, enabling communication for paralyzed individuals;
- BCIs, tiny devices placed on the brain’s outer layer, use electrodes to pick up real-time signals from neurons;
- The rapid development of AI and machine learning is accelerating progress in BCI technology.
This is an article from IO Next: The Brain. This magazine is full of stories about scientists, entrepreneurs, and innovations share one common goal: to better understand the most complex system there is.
One bright spot is that large parts of the brain often remain intact, even in the case of severe paralysis. So, there lies an opportunity for scientists, doctors, and health tech entrepreneurs. They can increasingly measure brain activity in detail and use this data to restore lost functions – such as speech – to patients. Technologies that measure changes in blood volume, magnetic fields, and electrical impulses due to brain activity are developing rapidly.
Communicating thanks to a four-four mm device
That technology comes in the form of a Brain-Computer Interface (BCI): a tiny four-by-four-millimeter device that can be placed on the wrinkled outer layer of the brain. Electrodes on these small “grids” pick up real-time signals from nearby neurons; a computer filters functional components and translates them into commands for an external computer. A BCI is nothing new; a patient was first fitted with this technology in 2015. However, the development of AI and machine learning makes developments go faster than scientists thought possible five years ago.
Neuro rights
As more and more commercial parties pick up BCI, technology is increasingly leaving the academic and hospital worlds. For example, Musk is working with Neuralink on a BCI with a thousand electrodes. Earlier this year, he was given the green light to test the BCI on humans.
It also raises an ethical debate about privacy and people’s identity. For example, Musk plans to link BCIs to smartphones. That would mean neurodata would be in the hands of tech companies. Therefore, a discussion about neural rights and regulations to protect people’s brain data from BCI misuse is critical. Chile and Spain, among others, are already working on a new constitution that specifically mentions neural rights.
From 0 to 100
“When I started in this research field 18 years ago, there was no technology to apply BCI in humans. Eight years ago, we successfully implemented a BCI in a patient. We want to introduce a fully implementable system with 128 electrodes in a few years,” says Erik Aarnoutse. He is an assistant professor at the Utrecht BCI Lab at UMC Utrecht, founded by Nick Ramsey in 2005. His career is dedicated to better understanding our brain and developing BCI technology.
The advanced BCI that Aarnoutse discusses, UMC Utrecht, is developing with the Graz University of Technology, Cortec Neuro (Germany), and the WYSS Center for Bio and Neuroengineering from Switzerland. Last year, the consortium received a six million euro grant from the European Innovation Council to develop ABILITY (Active Brain Implant Live Information Transfer sYstem). This is a neural recording system designed for home use. It can amplify low to high frequencies of brain activity and transmit it wirelessly to a computer.
The system can use electrodes to detect letters and words users say in their heads. Aarnoutse: “This also allows patients to initiate a conversation instead of always waiting for someone to look at their eyes. That makes a world of difference.”
A precursor to ABILITY (not wireless yet, but with cable) should be implemented this year. Implementing a BCI is complex: users often need to practice for an extended period before controlling the device. The BCI with 128 that Aarnoutse is working on should be able to do its job reliably within a few weeks.
You could compare this process to getting control of a new and unnatural body part. The BCI must also learn to extract the intention for movement or speech from brain activity.
Analyzing large amounts of brain data with AI
But, all the technology surrounding AI is developing at lightning speed. “AI algorithms can collect signals from the brain in large numbers. To train a BCI, the user has to say a certain word very often, ‘in the head’, while we record brain signals. This is how we train the AI algorithm. After this training, the algorithm can recognize the words without us knowing what the algorithm is paying attention to. We assume that there is a code in the electrical activity of brain cells. AI can decode those signals much faster than humans ever could,” Aarnoutse said.
For example, Radboud University and UMC Utrecht researchers converted brain signals into audible speech. By decoding signals from the brain using a combination of implants and AI, they could predict the words people wanted to say with an accuracy of 92 to 100 percent.
The researchers asked non-disabled people with temporary brain implants to pronounce some words aloud while their brain activity was measured. They directly translated brain activity on one side and speech activity on the other.
“AI can decipher signals much more efficiently and quickly than humans. Some say there has to be a more complex system than the brain before we can understand the brain. AI could be that more complex system.”
Erik Aarnoutse
A powerful and highly complex computer
The brain is the most complex human organ, with 86 billion brain cells making 10,000 connections with other cells on average. You can compare it to a potent and complex computer, the entire structure of which, to date, no one has been able to understand, successfully mimic, and simulate.
“Our brain is a gigantic network. When you learn something new or more efficient, connections are made. The complexity is in learning those connections and networks. But because there are so many, that is a huge task,” Aarnoutse said. He describes his career as a brain researcher as “very interesting, but at the same time also incredibly frustrating.”
The development of artificial intelligence is going to bring about a lot in brain research, Aarnoutse believes. “AI can decipher signals much more efficiently and quickly than humans. Some say there has to be a more complex system than the brain before we can understand the brain. AI could be that more complex system.”
