S A FRANCISCO – Okay, "the sewing machine" is pretty cool. But if the unit that Elon Musk's neuro-technology started Neuralink was developed to implant thousands of electrodes into brains (of rats and monkeys so far, and humans ultimately) was the only achievement, Tuesday's great revelation had been a great meh. Instead, six independent experts in the type of brain computer interface that Neuralink is developing told STAT they are most impressed, but there are surprises.
"In summary, the concept is impressive and it is the progress that they have made," says neurobiologist Andrew Schwartz at the University of Pittsburgh Medical Center, a pioneer in technology. "But much of this still seems to be conceptual. It is difficult to tell what is desirable and what they have actually done. "
The immediate goal of the San Francisco startup is a system that allows people who are paralyzed to use their thoughts to drive computers and smartphones. This has been done previously, including by Schwartz's group and one at Brown University where two tetraplegic patients implanted with the "BrainGate" neural interface system in 201
But despite decades of research, the systems have so many disadvantages that they are still not used in This is where Neuralink's technical reason can make a difference.
Dr. Leigh Hochberg from Brown University, who helped develop the brain computer interface for patients with tetraplegia, called Neuralinks "a new and exciting neurotechnology … I am happy to see how they will translate their system against the first clinical studies. "
An important feature of Neuralink's system is the large number of electrodes it plans to implant via its" sewing machine ", where a stiff needle quickly shoots thin-film polymers containing electrodes in the electrodes in the brain, in a white paper released on Tuesday night. by "Elon Musk & Neuralink", the company said it had implanted 3 072 electrodes in rat brains.
However, there appeared to be differences between what Musk and his team presented in their splashy revealing event in San Francisco on Tuesday night and what they said in The research document that has not yet been published or peer-reviewed and published Wednesday morning on a preprint server.
"The paper I would say is much less ambitious than the overall presentation," said Jacob Robinson, a neuroengineer at Rice University. 19659003] During the presentation, Musk and his team talked with excitement about a brain-computer link that could go in both directions – recording neural activity and also stimulating it. In the model, the interface should be able to stimulate the somatosensory cortex in a way that would cause paralyzed patients to feel as if they had touched something.
Researchers in Pittsburgh achieved it in 2016. But Neuralink's paper does not show a similar ability. That's an important omission, Robinson said, because "because the electrodes they work with, stimulation will be much more challenging than recording."
Musk in the presentation also spoke dreamily about using the technology to merge human brains with artificial intelligence. There is no information in the White Paper to support such a vision.
An Unresolved Longevity Problem
In his own research, Robinson is working on developing an interface to stimulate and record brain activity. He is particularly interested in "the actual electrodes that must live inside the tissue – and listen to brain activity and exist for a long time and hopefully do not screw things up," he said.
Therefore, Robinson was pleased with hearing Musk and his team referring to the field's still-unresolved longevity problem – the stubborn reality that at some point, whether slow or simultaneous, the electrodes implanted in the brain will stop working.
Such a breakdown would probably require patients to get a new implant – and to take the old one out. Removal can be particularly challenging for Neuralink, as it uses remarkably thin electrode-holding "wires," Robinson says.
There is also a safety risk associated with the faults of the electrodes, if this melting is caused by an infection likely to be in brain tissue or in the wire implanted in the brain. Robinson praised Neuralink's researchers for being caring about trying to dramatically reduce the risk by setting a goal to develop a device that is wirelessly driven through the skin to close the implant site.
But Neuralink doesn't have that technology yet. In the system described in the newspaper "there is actually a path of infection," says Robinson.  In addition to infection, safety risks from an interface between the brain machine may include stroke, aneurysm or immune response. 19659003] "The big risk is, of course, if something goes wrong … an additional major side effect can lead to a stricter regulatory thinking," said Andrew Hires, neurobiologist at the University of Southern California, whose research focuses on how brain processes feel the touch and feelings. in the cortex.
As a precautionary, Hires pointed to gene therapy, where an 18-year-old death in a clinical study in 1999 chilled investments and research for a decade. Still, Hires is optimistic about Neuralink's prospects and calls it the technology that it revealed "serious and credible".
A question about design
Neuralink says it has not found how many electrodes it would need for a brain-computer interface in patients, but it seems to have the ability to hover past existing systems. Prototypes tested in patients use hundreds of electrodes, while some are used for monkey research approaching 2000. "Neuropixels" developed by physicist Tim Harris of the Howard Hughes Medical Institute's Janelia Research Campus and colleagues have 960 electrodes per "shaft", a device that resembles the teeth of a comb, with electrodes on each tooth. Researchers routinely add four or eight eight in their lab rats, but Neuropixels are not yet used in human brain-computer interfaces yet.
The more electrodes, the more neurons the activity can monitor a system, including the "nails" indicating a neuron has been fired. The more neurons being monitored, the more information is fed into the software that analyzes their meaning and translates it into an output that "moves the robot arm 3 inches up and leaves and grabs what is there."
"With single-spike resolution and recording from more neurons, you can control more things more accurately," says Schwartz. With existing EEG systems, patients spend months thinking they must "now want to move my high toe" and then "I want to tilt my nose" to move a computer marker to the right-hand corner of a web page say. "If you could record individual nails [the brain-computer interface] it would work as soon as you hit it," he said. 19659003] Neuralink's electrode-holding threads are about 1.6 millimeters long according to the company's paper. "It would cover most of the cortex of a human," Harris says, and therefore reaches neurons beneath the surface that most such systems record. the geometry you need. "
Perhaps not surprisingly for a Musk company, the system's back-end electronics, which receives, amplifies, and decodes the brain activity detected by the implanted electrodes, experts more impressed. n The Brain Part The miniaturization, Harris said, is "substantially advanced over known technology".
The quality of the system's brain end is not as clear. According to data in the paper, no electrode neuron activity from rat brains detected as strong as 100 microvolts, which Harris calls the smallest signal "that you can rely on."
If he saw such weak signals from Neuropixels, he said "I think something is wrong. These are very low amplitude signals. What happened? If we had this data I would think something had gone south, like ours. probe missed its target or we hurt to insert it. "However, damage to the brain tissue can heal, but if it is the cause of the weak signals it may be temporary. Still, he said, "I look at the Neuralink paper and think, it's not what I want my data to be."
On the plus side: Neuralink's "sewing machine" can implant the wires that hold thousands of electrodes in the rat brain in one hour or less. It can take longer for people, but the best existing system, tested in rat brains. requires 48 hours of operation. "It's a big difference, driven by the robot that Neuralink built," says Harris.
On the minus side: Brains may not be friendly to have threads tied into them. "My guess is that the fast deposits make very long-term damage, so they may not get high-quality recordings for a very long time, but it's just a guess at this time," says neuroscience Loren Frank of the University of California, San Francisco. "We enter the brain much slower, and it's probably an important part of why we could get such long-lasting recordings" in rat experiments.
Challenges of money gambling cannot solve  Neuralink's promise to start testing their system with patients next year seems ambitious, if not misleading, to any expert consulted. A more realistic timeline, said Hires, may mean that Neuralink can start safety testing in humans within a few years – after it has produced data showing that it can safely implant its technology into non-human primates and that there are no long-term consequences for do it.
But, regardless of the time frame, Schwartz said: "I have the right concept and do what needs to be done."
With $ 158 million in funding – $ 100 million from Musk – Neuralink will have the luxury of graduates and less wealthy funded companies just dreaming about.
When Neuralink's researchers design their chips, they may try twice as many patterns as a typical academic lab can afford and help them move faster, Robinson said.
But throwing money around cannot overcome other obstacles.
Think about how long an electrode can live functionally inside the brain. "You cannot accelerate the process. You just have to wait – and see how long the electrodes are left. And if the goal is that these should be decades, it is difficult to imagine how to test this without waiting for long periods to see how good the devices work, "Robinson said.
The same can go to navigate the Food and Drug Administration.
"As much as I think with the right kind of resources, they can really drive technology at an incredibly fast pace," Robinson says, "the biological processes and regulatory approval … will be harder for them."