He Liping, chief reporter of The Paper
【Editor's Note】
The Chronicle of Cargo Colonization is the earliest historical work dedicated to the outstanding figures engaged in "cargo farming" (commercial) activities, and the economic ideas and commercial wisdom of the people expounded by Sima Qian are praised as "historical thought and economics, and the book cover is a pioneering work".
A new round of scientific and technological revolution and industrial transformation is reshaping the world economic structure and reconstructing the global innovation map. In this great change, the stories of all entrepreneurs, entrepreneurs and workers who dare to innovate and take responsibility deserve to be remembered. We launched the "Surging Financial People Weekly Magazine " To Tell the Story of Business People in the Tide of Globalization.
They have made a biography for the times, and we have made a biography for them.
The human brain has at least 100 billion neurons, about equal to the number of stars in the Milky Way. These neurons form 10^15 neural connections, connecting complex connected nerves end to end, with a total length of more than 180,000 kilometers. Such a complex network has given human beings intelligence, but the exploration of the brain has become the "ultimate frontier" that scientists around the world have worked hard to reach.
Implanting a tip-edge conductive, insulated electrode into brain tissue to closely monitor and decipher the neuron's "communication information", more ambitiously, hoping that the outside world and our brain will directly form a "two-way communication". These ideas date back to the 1920s, when a German doctor began to try to find EEG signals, which at the time looked like "black technology" and have now become an important part of brain science research in various countries.
In the first half of this year, the team led by Li Xiaojian, senior engineer and doctoral supervisor of the Institute of Brain Cognition and Brain Diseases of the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, and its cooperative team completed the long-term implantation of double arrays (more than 1400 channels of flexible electrode arrays) in the macaque brain, and successfully and continuously collected the neural signals of each channel with a high sampling rate of tens of thousands of hertz using self-developed instruments, taking the lead in opening up the whole technology chain of brain-computer interfaces in China. And they are also trying to use China's first self-developed super thousand-channel brain-computer interface system to carry out macaque two-handed collaborative movement brain-computer interface experiment, which is considered to be more promising than Neuralink's previous achievements in playing "MindPong" with macaque monkeys.
Recently, Li Xiaojian said in an exclusive interview with the surging news (www.thepaper.cn) reporter that the domestic brain-computer interface technology is not behind the international advanced experimental team in terms of various links, but how to organically combine the various ring technologies with high quality, that is, the so-called high-performance system integration, is the problem that needs to be overcome urgently. This is also the strength of Neuralink, a brain-computer interface technology startup of Silicon Valley "Iron Man" Elon Musk.
Li Xiaojian, senior engineer and doctoral supervisor of the Institute of Brain Cognition and Brain Diseases, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences. Courtesy of the interviewee
"If everyone works hard at this time node, there will be no obvious scientific and technological gap in brain-computer interface technology." Now is also the stage when neurotechnology begins to enter the market transformation, if we can also keep up, then we and the United States technology represented by Neuralink should not be much gap. Li Xiaojian believes that in this race, there is still a great possibility of running and even leading in the future.
Li Xiaojian has a flat head, wears black-rimmed glasses, wears a T-shirt, and is full of engineer style. He set the time for the interview in the afternoon, and he confessed that it was basically the beginning of his daily work, and his favorite was the late night work hour, which was the most efficient time of the day. When it comes to brain-computer interfaces, he is like having a huge bandwidth, and his professional vocabulary is like a cannon.
But unlike the engineer fan, Li Xiaojian is also good at expression, and can use very simple language to narrow the distance between esoteric science and the general audience. He likes to use the analogy of movies, in a speech, he said, "I believe everyone has seen the classic science fiction film 'The Matrix', in the film, the protagonists' brains are connected to the computer through a cable, they can directly swim in the virtual world with consciousness, for what they will not, it only takes a few seconds to input relevant knowledge into the brain to learn." In addition, in Cameron's "Avatar", the male protagonist controls Avatar through consciousness replacement, and then through Avatar and the dragon to carry out consciousness docking, with consciousness directly controlling the dragon. There is also the connection between the brain nerves and electronic devices in Alita: Battle Angel, which can directly control the entire mechanical body through consciousness. These are indeed very science fiction, and now with the development of brain-computer interface technology, is it possible for science fiction to become a reality? ”
Li Xiaojian has a bold prediction: brain-computer interfaces in the 6G era, human beings can use consciousness "group chat".
Xiaojian Li graduated from the Department of Chemical Engineering of Northwest University in 2001 and obtained his Ph.D. from the Institute of Biophysics of the Chinese Academy of Sciences in 2010. For the next 8 years, he conducted postdoctoral training and served as a research assistant at the Georgia Medical College and Northwestern University in the United States. "More dramatic", Li Xiaojian commented on his academic growth background: from the beginning of "using synthetic biological methods to make proteins, focusing on creating new life forms" to "inspired by brain science, researching more advanced forms of artificial intelligence", and finally trying to "integrate the human brain with computers".
In the autumn of 2018, Li Xiaojian returned to China, did not choose the academic circle closer to his hometown Tianjin, but went south to join the Institute of Brain Cognition and Brain Diseases of the Shenzhen Institute of Advanced Technology of the Chinese Academy of Sciences, which coincided with the launch of the Guangdong Provincial Brain and Brain-like Key Research and Development Program. Subsequently, Li Xiaojian participated in the provincial brain-like intelligence key technology and system research, and the national brain-computer fusion brain information cognition key technology research and other projects.
Li Xiaojian's current research areas are mainly in high-performance brain-computer interface and brain-like engineering, and the team focuses on implantable brain-computer interface research, including neural electronics and neurophoton technology for broadband brain-computer interface, neural circuit analysis and decoding technology for neural simulation and brain-like computing, and neural mimesis equipment and system research and development for humanoid robots.
"Brain-computer interfaces have actually been studied since the 1960s, and for a long time since then they have been in the exploratory and functional verification phase. Scientists have found that the animal's action behavior can be decoded from the neural electrical signals obtained in the brain, which is of course the neurophysiological basis of behavior, but it also shows that the brain-computer interface is feasible, and it has a scientific basis. Li Xiaojian said.
At present, the hot word - brain-computer interface, the interpretation is different, narrowly refers to the establishment of a direct communication method that does not depend on the five senses and limbs, and the brain and external electronic devices such as computers. And it has two English origin words, customary "Brain Computer Interface" mainly refers to the non-invasive brain-computer interface technology based on brain signal acquisition outside the brain, while "Brain Machine Interface" mainly refers to the brain-computer interface technology (that is, implantable brain-computer interface technology) implanted in the brain with neural activity sensors. Scientists hope that through such technologies, the brain can control external devices, and also hope to input information directly into the brain to achieve a two-way closed-loop information system.
It is worth noting that China's "brain plan", that is, "brain science and brain-like research" as a "major project of scientific and technological innovation 2030" will be fully launched. Tao Hu, deputy director and researcher of the Shanghai Institute of Microsystem and Information Technology of the Chinese Academy of Sciences, previously wrote in an article that with the advancement of the plan, the analysis of brain cognitive principles, the pathogenesis and intervention technology of heavy brain diseases related to cognitive impairment, brain-like computing and brain-computer intelligence technology and application, children and adolescents brain intellectual development research, and technology platform construction will make great progress. Among them, the brain-computer interface, as the underlying core technology, is related to almost all the key contents of China's "brain plan".
Looking around the world, in addition to Musk, which has attracted the most attention, and neuralink, commercial giants such as the U.S. Defense Advanced Research Projects Agency (DARPA), Facebook, Google, and Amazon are actively laying out the field of brain-computer interfaces. Tao Hu reminded that at present, the key devices and high-end equipment of China's brain-computer interface (especially the implantable brain-computer interface) rely heavily on imports, and the domestic lack of original brain-computer interface core technology, most of the tracking, the layout is scattered, and the lack of systematicness. In the past two years, the United States has carried out export controls on brain-computer interfaces, and the supply of system-level products and core devices has been greatly affected, which has had different degrees of impact on China's brain science research and the treatment of patients with neurological diseases.
However, Tao Hu also believes that in high-end technology, brain-computer interfaces are one of the areas where China is most likely to catch up or even "overtake in a straight line".
From 4 channels to several thousand channels, smaller devices collect more nerve signals
Humans now know that nerve cells use electrical signals to communicate with each other, and this electrical signal is a short discharge. The amplitude of the brain nerve electrical signal is very weak, generally at the microvolt (uv) level, and it also has to go through the attenuation of the skull and scalp, and it needs to go through a thousand times magnification display and a filter that reduces interference to form an EEG that is not unfamiliar to the scalp now.
Measurements of EEG can be traced back to the 1920s, when the world's first EEG recording was made by Dr. Hans Berger, a German psychiatrist, who demonstrated for the first time that electrodes placed on the scalp of the brain could measure currents that reflected brain activity. Scientists' in-depth understanding of electrical signals in the brain continues to span nearly half a century.
It wasn't until the 1960s and 1970s that brain-computer interface technology really began to take shape. Until the end of the last century, it developed rather slowly. Li Xiaojian attributed the main reason to the backwardness of microelectronics and micromachining technology. He cites neural signal acquisition as an example: "In the early days, we could only use a single-root sensor to probe neural signals in the brain of an animal, basically only one or two neurons could be recorded, which was not enough for us to understand the brain." ”
The sensor used by Li Xiaojian in his early experiments was a glass-coated tungsten wire electrode, "a needle goes down, the front tip contact can collect nerve electrical signals, and the needle shaft part is insulator." An experiment cannot implant a few electrodes in the head, nor can it collect many nerve electrical signals." "Of course, in the 2000s, the computing power of the desktop computers that we analyzed data may not be as good as today's smartphones, and the data collected cannot be analyzed."
Because of this, Li Xiaojian joined the laboratory of Professor Qian Zhuo of the University of Georgia Medical University, an advocate of the "Brain Deciphering Project" after graduating with a doctorate. Known as the "father of the smart mouse," Qian Zhuo's lab had the highest device in the world to collect neural signals— a 1,024-channel neuroelergeological signal acquisition system developed and produced by Plexon. "At that time, the electrophysiological signal acquisition system was also quite large, basically the size of a cabinet. At full capacity, eight desktop computers are required to run at the same time. ”
"Similar to the history of the development of electronic computers, the system was huge at the beginning, and the circuit integration was not high." Li Xiaojian said: "When I was a graduate student doing experiments, the signal acquisition instrument I used was the size of a desktop computer, but it could only support 4 signal channels, and now our palm-sized equipment supports thousands of channels. He emphasized that the development of microelectronics and electronic engineering technology has greatly improved the degree of functional integration, allowing us to collect more neural signals in smaller devices.
Nowadays, Neuralink, which is the most popular, has made a lot of innovations in this regard. In July 2019, Neuralink announced its results for the first time – a scalable, high-bandwidth brain-computer interface system. The system consists of a set of electronic chips and silk threads that are only 4 to 6 microns thick and thinner than a human hair. The entire system contains 3072 sensors distributed on approximately 100 flexible wire threads.
In general, according to the technical way of obtaining brain information, the brain-computer interface technology at this stage mainly includes the traditional non-invasive scalp electroencephalogram (EEG, which is also the most mature technology), non-invasive imaging technology, and invasive brain implanted electrode recording electrical signal technology. In terms of spatio-temporal accuracy, intrusive has incomparable advantages, so even in the face of more safety and ethical risks, this method has undoubtedly become the first choice of the world's top brain-computer interface research team.
As far as implantable brain-computer interface technology is concerned, what challenges and difficulties still exist at present? Ni (National Instruments), the world's leading test and measurement manufacturer and has been working in the brain science or life sciences industry for many years, has observed this.
Guo Qiao, senior technical marketing manager of NI Asia Pacific, previously told the surging news (www.thepaper.cn) reporter that from the signal acquisition link, the key technical links and their difficulties are as follows: there are further optimization needs for miniaturization, special material selection, biocompatibility, rejection reaction, etc. in the electrode; front-end signal conditioning and amplification (filtering, noise reduction, etc.); multi-channel synchronous signal acquisition, that is, hundreds to thousands of channels of synchronous acquisition, high dynamic range, 16 bit/24 bit ADC bit width, sampling rate of tens of thousands of sampling points per second; multi-channel signal real-time processing, ultra-high channel acquisition of digital signals need to be through software-defined hardware (such as FPGAs), convenient for researchers to quickly achieve neural signal filtering, denoising, decoding, analysis and other innovative algorithm prototype implementation and verification; closed-loop feedback system.
Guo Qiao mentioned that the first three links of China's major domestic brain-computer technology-related universities, scientific research institutions and companies are in the process of tackling key problems, "The focus of the NI platform is the last two links, and self-developed or ready-made acquisition systems are used to build a complete closed-loop real-time system." ”
Is the lack of understanding of brain science a barrier to brain-computer interface?
The more than 3,000 electrodes that Neuralink's sewing machine can implant are clearly not the end point. Billions of nerve cells are often involved in brain activity, and tens of thousands of electrodes may not be enough for scientists.
"Some people have proposed to do a million-channel brain-computer interface system, of course, some people have mentioned that it may require hundreds of millions of channels, such an idea is mentioned, but from a technical point of view, it is not possible to achieve it in a few years, of course, some people think, is this idea too rough?" Li Xiaojian mentioned that before implanting more electrodes, what needs to be explored is: What information do we need the most? Where does this important information come from?
In his view, it was more feasible at this stage to conduct research on specific tasks. In the case of non-invasive functional magnetic resonance imaging (fMRI), for example, it can capture an overall picture of the whole brain, but it cannot give detailed information in both space and time. In exchange for neuroelectronic signals, its spatiotemporal resolution is more than a thousand times higher, "but there is a problem here, our current brain-computer interface technology can not achieve full brain coverage, can only be implanted in a few specific locations, and there is a certain limit to the number of sensors implanted in the brain region." "The increase in the number of implanted electrodes also means increased surgical risks and ethical controversies.
How can brain-computer interfaces enable proficiency in multitasking rather than performing a single task as currently demonstrated? This high-performance requirement for brain control makes brain-computer interface research face the challenges of increasing the amount of information collected and further deepening the cognition of brain science. "How many brain regions does it involve?" How much information is collected from these brain regions? Can today's technology support it? That's what we're doing now. Li Xiaojian expects that in general, it may take 5,000-10,000 collection fluxes to basically achieve multi-tasking functions.
In Li Xiaojian's view, there is also a problem of "two-way difficulty" between brain science and brain-computer interface. "The basic research of brain science is not very clear, which limits the rapid expansion of brain-computer interface application scenarios; and brain-computer interface technology itself is also an important tool for brain science research, and it is urgent to upgrade technology to accelerate the in-depth research of brain science." He believes that brain-computer interface technology can not only push its application scenarios in the medical field, but also promote its value as a high-performance tool for exploring the operation mechanism of the brain and inspiring the design of brain-like intelligent systems.
After the massive amount of brain nerve electrical information is collected, how to deal with it quickly? This is another problem. "The brain-computer interface needs to be decoded in real time, it needs to be collected very quickly, and it needs to be processed very quickly, which is a basic premise and should be completed within tens of milliseconds."
Guo Qiao also told the surging news reporter that the ultra-high channel plus 16bit/24bit data of tens of thousands of sample points per channel per second has very high requirements for system bandwidth, and the data on the G bandwidth needs to enter the FPGA board for real-time processing. "Entering the FPGA board, neural signal filtering, denoising, decoding, and even AI deep learning algorithms require a large number of fixed-point number operations, including signal operations between channels, and real-time processing usually requires operations to be completed in milliseconds or even subtleties."
In the processing of brain nerve information, Li Xiaojian believes that it is more suitable to look at the two routes separately at present. One route refers to the need to continue in-depth research after the neural signal acquisition, "our understanding of the brain is not particularly clear, the signal must have scientific research purposes, this aspect of the use of course hope that the more detailed the neural signal collection, the better, and then store it, and then can be more detailed analysis through the supercomputer." ”
Another route is used in the actual application scenario of the brain-computer interface, "We need to output the decoding result at the fastest speed, which requires meaningful data compression, obtaining the neural information that we really want to use and use, processing it in near real time, and outputting the decoding result." ”
Depending on the two paths described above, the respective technician will have a different strategy. Li Xiaojian once again took The example of Neuralink's previous achievements, "The wired brain-computer interface system they showed with rats in 2019 has more than 3,000 channels, but the number of channels of the wireless brain-computer interface system shown in the last two times is much lower, and in fact, a lot of data compression has been done, and the data transmitted wirelessly is very small." Since the corresponding task is simple, this little bit of information is enough for the monkey to control the movement of the light spot on the screen with his mind. ”
For another key ring of brain nerve electrical information decoding, scientists are still solving the real decoding method.
Pu Muming, an academician of the Chinese Academy of Sciences and the leader of China's "Brain Plan", pointed out at the 2021 Pujiang Innovation Forum that there are still great obstacles to the field of brain-computer interface, and some major scientific problems need to be solved. The first big question is how to decode the brain information? He believes that the existing collection of brainwave information, using artificial intelligence algorithms and machine learning algorithms to decode the information represented by the brain's brainwaves, "this is a superficial decoding, not a real understanding of brain activity." ”
Pu Muming explained to the surging news (www.thepaper.cn) reporter that now it mainly relies on big data, and the data is not precise enough, "I don't know what it means, I have to correspond." He thinks the future of EEG should be more refined and more data, "but if you know what the loop looks like, it's even better." You know that this loop represents this finger movement, that loop represents that finger movement, and you record it above, and you know very well that you don't have this ability yet. ”
Speaking of this problem, Li Xiaojian believes that the brain is divided into multiple brain regions because of its different functions and different ways of processing information. "In the primary cortical region that processes continuous dynamic information, or the brain region that produces a discrete abstract information structure, there can be a corresponding decoding method. But in many intermediate, jointly functional brain regions, the information is still in a mixed state, and we don't have a good way to extract the information here effectively. What is the information that corresponds to our cognitive tasks? And how do you decode it? There are still great challenges. ”
In May this year, the top academic journal Nature published a landmark study jointly completed by researchers from Stanford University, Brown University, Harvard Medical School and other teams in the form of a cover article, combining artificial intelligence software with a brain-computer interface device to allow a paralyzed patient with a brain-computer interface device implanted in the brain to imagine holding a pen and "trying" to write on a horizontal paper, as if his hand was not paralyzed. And the man's handwritten intent was quickly converted into text on a computer screen.
The study's corresponding author, Dr. Francis R. Willett, a research scientist at Stanford University's Neuroreparation and Transformational Laboratory, said in an interview with www.thepaper.cn, "When you can only record a small number of neurons collected by the sensor, it is helpful to have very different neural patterns, and the chance of accidentally confusing them will be low." That's why complex movements, such as writing different letters, may be easier to decode, and complexity makes them more unique and different from each other. Willett further explains that in contrast, the most advanced typing method before, "moving along a straight line to different keys evokes very similar patterns of neural activity, because all that is involved is a linear motion with different angles or distances." ”
Willett mentions that this also means, perhaps contrary to what we intuitively think, that decoding complex behaviors is more advantageous than simple behaviors, especially in classification tasks.
Full technology chain integration is the key, commercialization has come?
Since returning to China, the team led by Li Xiaojian has made a series of breakthroughs in the field of brain-computer interface.
The 1,000-channel neurophysiological signal resolution system developed by the team is small in size and has real-time parallel neural solving function, which is a broadband brain-computer interface electronic device of the same level as the Neuralink system. In collaboration with the Brothers Institute of the Chinese Academy of Sciences, they successfully implanted a double array of 256 flexible neural electrode arrays in the brains of rabbits and crab-eating monkeys. A few months ago, it successfully completed the long-term implantation of the double array of more than 1400 channels of flexible electrode array in the macaque brain, and used self-developed instruments to achieve high-speed synchronous collection of neural electrical signals in all channels, taking the lead in opening up the whole technology chain of brain-computer interface in China.
Prototype of the super two-thousand-channel cranial nerve signal collector.
At the same time, Li Xiaojian and others are also accelerating the realization of China's first macaque two-handed collaborative brain-computer interface experiment using the self-developed super-thousand-channel brain-computer interface technology equipment. The study will lay the foundation for China's independent development of an implantable brain-computer interface system for paralyzed patients. Paralyzed patients will regain their freedom of movement through brain-controlled mechanical limbs.
What is the team's level in the brain-computer interface field now? In response to this problem, Li Xiaojian modestly self-evaluated: "If you talk about each technical link alone, it is actually average. He believes that the significance of his past work is mainly that "we have had many years of systematic research on brain science and neural decoding, and now through self-research and temporary import of spare parts, we have built the entire system that we think is ideal, so that it can operate more efficiently and realize the penetration of the whole technology chain." In this way, we have a complete foundation for independently developing brain-computer interface technology and applications with our own characteristics based on our own thoughts, our own technology, and our own systems. At present, the team's brain signal collector can be only the size of a palm, achieving more than 2,000 acquisition and processing fluxes, and a bandwidth of more than 100 megabytes per second.
Macaque upper limb motor brain-computer interface model and experimental demonstration.
The benchmark that Li Xiaojian repeatedly mentioned is Neuralink. "From our point of view, all of their stuff is not a counter-natural technology, And it's not unusual for Musk to recruit experts in various fields, and the technology that these experts bring has been reported in top journals or conferences in various fields." The key advantage is the simultaneous optimal integration of the entire technology chain.
"Brain-computer interface technology spans multiple disciplines, how to carry out interdisciplinary research and development, how to bring multidisciplinary talents together and integrate them, this is actually the real frontier of the current brain-computer interface." Li Xiaojian further emphasized that the domestic experience in large system integration is not sufficient, "now it is equivalent to every technical link has something, but also do a good job, but the organic combination of these several technical points to exert the strongest efficiency, this is a kind of ability and process." ”
Guo Qiao also talked about similar pain points: the traditional brain-computer interface research team does not have or is proficient in all aspects, and how to quickly transform and implement scientific research results is a challenge in this field. What is quite inconvenient is that the current commercial or ready-made systems are relatively closed, and it is difficult for researchers to make customized modifications or integrate their own innovative algorithms into the system for iterative development. "NI is committed to forming a complete and open ecosystem, whether it is docking third-party hardware or software algorithm IP, which can greatly reduce the development time of researchers, allow researchers to focus on their own areas of expertise, and accelerate the process of basic research on brain mechanisms."
It is worth mentioning that Li Xiaojian's Shenzhen Advanced Institute of the Chinese Academy of Sciences has quite an advantage in interdisciplinary cooperation, and its positioning is also to become a new type of international first-class industrial research institute. Li Xiaojian said, "Brain-computer interface has both basic and applied research, and needs a multidisciplinary and interdisciplinary integrated team of researchers including basic brain science, neural engineering, mechanical and automation, electronics and computers. In this regard, Shenzhen Advanced Institute has a natural advantage, and project cooperation does not even need to go out of the compound. For example, there are Dai Ji, an associate researcher in charge of the non-human primate experimental platform, Deng Chunshan, a senior engineer engaged in sensor design and evaluation, Zhang Danke, an associate researcher engaged in neural decoder algorithms, Du Zhanhong, an associate researcher engaged in biocompatibility enhancement, and Wang Yishan, a senior engineer in electronic chip design.
Of course, in actual research, Li Xiaojian and others will also "walk out of the compound." "Other research institutes, including our brother firms, have many years of deep cultivation and reserves in their own research fields, and we are also working closely with teams that have a great advantage in a certain technology."
Team members
Compared with its benchmark Neuralink, Li Xiaojian also talked about some gaps, such as promoting higher throughput sensors and having better tissue compatibility and service life, promoting the self-development and self-production of neuroelectronic chips, and more integrated systems to achieve complete in vivo implantation. "After all, we are still in the prototype stage, although it is the size of a slap, it is still much larger than the Neuralink device, and we hope to further miniaturize."
In fact, to expand to later applications, miniaturization is a crucial step. The miniaturization of devices and the wirelessization of communication will directly affect the freedom of movement of implanters and determine the future popularity of the technology. In the vision of Li Xiaojian and others, the brain-computer interface that enters practical application in the future should have the characteristics of "four modernizations", that is, minimally invasive, miniaturized, wireless, and brain-like intelligence.
How does the technology end up? Musk mentioned at the beginning of the establishment of Neuralink that its short-term goal is to treat brain diseases such as Alzheimer's disease and Parkinson's disease, and its long-term plan is to enhance the function of the human brain by implanting some "chips and cables" to fuse the human brain with AI.
"The human brain to integrate with artificial intelligence are all visions, it will take decades, the most optimistic estimate is also 20 years, but in fact, what can be done in a few years is actually mainly medical use, for patients with various brain diseases." This is exactly what Li Xiaojian himself wants to promote at present.
Does Neuralink's popularity also mean that the "tipping point" for the commercialization of brain-computer interfaces has arrived? Li Xiaojian said that the implantable brain-computer interface belongs to the invasive type, even if it is minimally invasive, the corresponding products also belong to the three types of medical devices. The so-called three types of medical devices refer to medical devices implanted in the human body to support and maintain life, which are potentially dangerous to the human body and must be strictly controlled for their safety and effectiveness. "That means approvals are cumbersome."
In August 2020, Musk announced that Neuralink's brain-computer interface devices had been certified by the FDA's Breakthrough Devices Program in July of that year, and that implantation experiments would be performed on humans, and the company was planning to make more experimental approval declarations. Another competitor, Synchron, founded in 2017, was approved by the FDA in July to take the lead in conducting human clinical trials of its product to verify the safety and efficacy of its flagship product, Stentrode motor nerve prosthesis, in patients with severe paralysis.
To distinguish, Synchron uses a method similar to cerebrovascular stents, which delivers a network of Stentrode sensors to the brain through blood vessels in a minimally invasive manner. "This method is easier to approve clinical trials, but the electrodes can only be parked in larger blood vessels. There are few available sites and long distances from neurons, and a large amount of brain nerve information cannot be clearly obtained, so this is still two different things. Li Xiaojian always believes that the approval of future clinical trials of Neuralink brain-computer interfaces can finally drive the "opening of the floodgates" in this field.
Before the real "opening of the floodgates", what kind of pace should the domestic team maintain? "If everyone is working hard at this node, there will be no obvious scientific and technological gap in brain-computer interface technology." Now is also the stage when neurotechnology begins to enter the market transformation, if the brain-computer interface can also keep up, then we and the United States technology represented by Neuralink should not form much gap. Li Xiaojian said.
Previously, Pu Muming stressed to the surging news (www.thepaper.cn) reporter that the goal of the Chinese team is not limited to catching up with Musk's technology, "the direction is different, our vision is bigger, and it takes a lot of effort to do a closed-loop brain-computer interface." ”
Editor-in-Charge: Li Yuequn
Proofreader: Yijia Xu