“Human computer interaction”, “FDA breakthrough equipment certification”, “Fitbit in the brain”, “human experiment approval”, “human brain chip installation”, “accurate prediction of experimental pig movement trajectory”… On August 29, when musk announced the latest progress of brain computer interface technology under neuralink, his brain computer science company, these words instantly burst the circle of friends.
As one of the important products of Musk’s early layout in the medical field, he took over neuralink, a company dedicated to developing “neural network” technology in 2017. Since then, although neuralink has experienced personnel turnover for several times, new topics flow out every year. From last year’s “sewing machine like” robot to this year’s coin sized brain computer chip, musk will cause a carnival every time he releases information about the brain computer interface.
Musk once again linked BCI to mental illness. He mentioned in his speech that many people may encounter various neurological problems at different stages in their lives, such as amnesia, blindness, deafness, paralysis, depression, insomnia, addiction, epilepsy, stroke, brain injury, etc. “The value of neuralink is to provide an affordable and reliable solution to these troublesome problems. It has been proved to be feasible to solve these problems by implanting electronic devices into the brain.”
However, the blueprint is beautiful, but the reality is as musk thought?
According to overseas media reports, musk developed brain computer interface out of fear of artificial intelligence. He believes that according to the current development speed of AI, it will not be long before human beings will be ruled by AI and become puppets. Therefore, he hopes to find a way for humans to fight AI. The only way is to make human beings stronger, so he put forward a more “sci-fi” idea than immigrating to Mars – brain strengthening. Musk once said: “human beings need to integrate with machines and become ‘semi robots’ in order to avoid being eliminated in the era of artificial intelligence (AI).”
However, the actual products are indeed placed here. Is BCI technology the carnival of cyberpunk lovers or the Savior of neurotic patients? Arterial network interviewed many enterprises and experts, and integrated their views here, hoping to describe the development future of BCI from a neutral perspective.
Algorithmic challenge: 4 steps to apply BCI to practice
According to Musk’s logic, the ideal brain computer interface can not only help researchers collect neuron signals, but also encode specific instructions, transmit them to other parts through the brain computer interface, and assist the brain to complete signal transmission.
Then, to realize this process, at least four processes need to be completed: signal acquisition – signal decoding – recoding – feedback.
These four processes seem simple, but in fact they are extremely difficult. Just the first “signal acquisition” process has blocked a large number of explorers who want to dig gold from BCI.
Neuroscientists often use stadium analogy to describe the process of brain signal acquisition: outside the stadium, you may hear background noise and judge whether a team has scored from the cheers; When you sit on the top of the court, you can know which team has scored this point; But only when you sit close enough and understand the cooperation logic of football, can you know how a set of coordinated actions helped the team score the goal.
This is also one of the important reasons why neuralink’s BCI has developed from a “sewing machine” to today’s “intrusive coin” – only when the electrode network is close enough to the neuron can we obtain a sufficiently high-resolution signal.
It can be seen from the pig video released by musk that its implanted electrode has indeed solved this problem. During the demonstration, the staff read in real time and synchronously displayed the brain waves of pig B on the large screen. Neuralink, installed in pig a’s head, is reading the current on the nerves related to its nose. Whenever its nose touches anything, there will be a peak of brain waves.
In the video of the second pig on the treadmill, he demonstrated the use of brain waves to predict the trajectory. The chart shows that the predicted trajectory is basically consistent with the real trajectory.
Neuralink has been able to predict the piglet’s posture to a certain extent, which means that once the collected signals reach a high accuracy
However, although musk has made a major breakthrough in signal acquisition, we do not seem to see much breakthrough in the second stage of BCI implementation – signal decoding stage.
“What disappointed people at this press conference was that there was no progress in neural signal decoding. It simply demonstrated the relationship between the movements of the piglet’s limbs and the nerve discharge in the brain. There is still a long way to go before the brain computer interface is implanted and the mobile phone communication.” Professor Hongbo, an expert on brain computer interface at Tsinghua University, said: “at present, the research on brain computer interface decoding of motion information has been very mature. Brown University and Stanford University in the United States have successfully demonstrated on the brains of monkeys and humans for many times. However, the FDA in the United States has approved cyberkinetics and BlackRock to carry out small-scale human clinical trials in the past, but they have not achieved the expected results.”
“At the same time, this kind of research has also been carried out in China, mainly because Zhejiang University and Tsinghua University are engaged in relevant research. Zhejiang University uses the American Utah electrode array mentioned by musk in his speech, which has been implanted into the cerebral cortex of monkeys and patients, and successfully realized the control of the brain computer interface of the manipulator. Tsinghua University, in cooperation with 301 Hospital and Xuanwu Hospital, has carried out minimally invasive brain computer interface implantation on epileptic patients Different schemes are adopted in oral study. The recording electrode is buried in the skull without penetrating the dura mater, so it does not destroy nerve cells. It can collect intracranial EEG stably for a long time. The brain computer interface typing has been realized.
“It should be noted that these two research groups are still in the pre clinical trial stage and have not obtained the medical device license. The main technical bottlenecks are the same as those encountered by neuralink team, such as wireless transmission of nerve signals, control of nerve cell trauma, long-term safety and effectiveness of implanted electrodes, etc.” Professor Hongbo explained to the arterial network.
Then, if musk can complete the decoding problem in the next work, the encoding process in step 3 may not be so difficult. However, the feedback process in step 4 will inevitably be another insurmountable mountain.
The feedback link is to use BCI to obtain environmental feedback information and then act on the brain. Generally speaking, we rely on vision, hearing, touch and hearing to obtain environmental information and then transmit it to the brain in real time. However, even the current popular computer vision technology, which has been widely used in daily life, mostly stays in the processing of two-dimensional effects. The large amount of three-dimensional image data, difficult to encode and other problems have become a huge obstacle in the feedback process.
Therefore, from this conference, we really should revel in Musk’s successful acquisition of high-resolution neuron signals. Accurate and high-resolution signals can greatly promote the progress of decoding. But we also need to be rational. BCI algorithm is only one of the problems, and signal acquisition is only a part of the algorithm problems – musk still has a long way to go from mature BCI.
Material challenge: finding implants that can be retained in the intracranial environment
Different from other human environments, Professor claudeclement of Weiss center in the United States compares the brain to the coastal jungle: humid, hot and salty. “This is by no means a good place for technology.”
Different from the oral, intestinal, abdominal and other environments, the structure of the human brain is more sophisticated, full of mystery, and even carries the “soul”. It is not easy to place a sensor in this place. We need to consider both the rejection response of the brain environment and the durability of the implant to avoid frequent replacement and unexplained intracranial damage. The chip of implant also needs high technology. On the one hand, it needs to meet the collection and processing of millions of neuron information; On the other hand, it also needs to be small enough to avoid compression damage to other tissues in the brain. It can be seen for various reasons that the design of implants also requires the enterprises to consider carefully.
The economist’s “implant” once described two directions for implant design. The first is to rethink the current small conductive electrode technology, and the second is to move towards a new non electrical direction.
Professor Ken Shepard from the Department of electrical and biomedical engineering of Columbia University has achieved this in CMOS (complementary metal oxide semiconductor) electronic technology. He believed that any plug-in electrode could cause cell damage, so he tried to develop an integrated device placed on the top of the cortex and under the membrane surrounding the brain. In 2018, the first generation CMOS chip prototype he designed was only 1cm2, containing 65000 electrodes, while the second generation version will contain 1million electrodes. It is worth noting that instead of stacking the sensors on the chip, he added the same number of amplifiers to convert the signals and a wireless link to send data to the repeater on the scalp.
At that time, the chip did not solve the power supply problem. After all, it was difficult to ensure safety by placing devices such as batteries containing many dangerous chemicals in the brain. However, from the press conference, musk seems to go further in this regard. However, the press conference only said that the device has wireless charging function, but did not explain how to realize this function. In past speeches, neuralink has said that the battery of its implant can last 24 hours and can be charged wirelessly like a mobile phone. We can’t verify this from the video alone.
The video only shows the style and size of the implant
Returning to the topic of implants, from the non electrical direction, Dr. hongguosong of Harvard University tried to manufacture a porous network made of SU-8 flexible polymer embedded with sensors and conductive metals. This reticular structure mimics the elastic and soft shape of nerve tissue and allows neurons and other types of cells to grow in it, which means that it can solve the brain’s immune response to foreign bodies. Compared with traditional solutions, this solution blurs the boundary between biology and electronics.
In addition, recently, the joint team of the medical school of Tsinghua University and the Department of microelectronics adopted a new memristor array to try to process the signal of the brain computer interface, reducing the power consumption by 400 times, which is also a promising direction to solve the above problems.
In addition to the above three examples, many scholars have conducted in-depth research in the field of implants, which will not be repeated here. However, from the perspective of both algorithms and materials, musk still hasn’t solved the key problems of BCI.
According to the analysis of Bloomberg, neuralink uses flexible polymers, which are difficult to exist in the human body for 10 years. Dai Juyi, CEO of niunuo technology, a domestic brain science and medical overall solution provider, told aoji.com: “objectively speaking, Musk’s method is still within the scope of traditional technology, which can reduce the cost of existing technology, but there is still a long way from ‘treatment’.”
Ethical problems: ethical review of human experiments cannot be avoided
Although musk said at the publisher that the BCI product has obtained the breakthrough device certification granted by FDA in July this year and will be further put into human clinical trials, after all, since the product is only in the primary stage, the risk of human implantation is relatively small. However, with the deepening of technology, there are serious human ethics problems here. In the discussion group of cyberpunk enthusiasts, the topics of control and being controlled, military war and so on triggered by the BCI end may not get a relatively consistent result for several years.
This is also true in practice, especially in the feedback link that BCI needs to face eventually. A scholar from Harvard University once said: “if BCI is mature, researchers can safely put a person into or out of the ‘locked’ state – a person can be fully aware of his environment, but can not make any action or action. However, such an experiment cannot be tolerated by the modern audit committee.”
Professor Hongbo also questioned the ethical issues of BCI. He believes that the ethical issues of BCI implantation and gene editing are equally important, and the corresponding ethical framework should be established at the same time of technology development, rather than making up for the lost sheep.
From the current development form, in order to better explain ethical issues to managers, researchers had better explore the integration of BCI and brain, so that they can guide people’s actions through collaboration. Such collaboration must be explicable. Perhaps such a system can manage to deal with ethical issues, but it must still face thorny review, let alone clinical trials.
A thought-provoking question can be raised here – is it possible for China to catch up with musk in BCI technology and deal with the ethical issues? Unfortunately, the answer to both is No.
First, let’s talk about the second problem. At present, China has not formulated a special approval procedure similar to FDA for “breakthrough equipment” to help BCI and other products obtain approval. In contrast, the green approval channel of China’s equipment audit center emphasizes the changes in economic benefits and curative effects in the “treatment direction” brought by innovative equipment, and the BCI designed by musk is obviously not in this category. Secondly, China has not been able to catch up with Musk’s technical achievements in the field of BCI for a long time, but this does not mean that China has ignored the development of brain science. In fact, in the five years from 2015 to today, China has surpassed the United States in the clinical research of many neurological diseases.
However, from a macro point of view, basic science such as BCI is obviously crucial to human development. It is precisely because a large number of scholars have invested and dedicated themselves in basic science that we now enjoy science, technology and life. Therefore, Musk’s greatness is self-evident.
How should China develop brain science in comparison with Europe and America?
As mentioned above, although the domestic BCI technology cannot be compared with that of the United States, our brain science research level is still in the forefront of the world.
At the same time, based on the advantage of population, we will get more big data of brain diseases over time, which will help more patients to be cured – BCI is not the only way to cure brain diseases.
So, how should China promote the development of brain science? The integration of direction, talent and industry may be the solution.
Different from Musk’s grand blueprint, domestic brain research pays more attention to the diagnosis and treatment of brain diseases and brain inspired artificial intelligence. Around 2019, after more than five years of gestation, China has formed a basic consensus in the scientific community on the research direction of the “one body, two wings” brain program.
“Integration” means “cognitive brain”, which focuses on and understands how the cognitive function of human brain comes from. The core is the essence of cognitive brain structure and functional neural network, trying to clarify how the brain works.
Academician pumuming, the leader and advocate of the Chinese brain program, said: “when we see a computer, we must know the structure of the computer to analyze its functions. For the function of the brain, we must know the network structure of the brain. This is called the ‘whole brain mesoscopic neural connection diagram’, which is also a key part of our big program.”.
The two wings point to the two main attack directions of “protecting the brain” and “creating the brain”.
Among them, “brain protection” is mainly to better diagnose and treat all kinds of major brain diseases, including Alzheimer’s disease, epilepsy, Parkinson’s disease, depression and other diseases. On the race track of nervous system diseases, there will be an opportunity to produce 100 billion unicorns.
“Creating brain” mainly realizes the research and development of brain like artificial intelligence. The core strategic goal is to develop brain like computers, which will consist of two parts: one is to develop brain devices and structures; The second is the design and development of brain information generation and processing system.
The great value of the Chinese brain program lies in its continuous implementation in the next five to ten years. It will spare no effort to promote the deep integration and development of artificial intelligence and brain science. Its research results will greatly promote the development of brain like artificial intelligence technology. Research breakthroughs in this field will lead to a new round of scientific and technological revolution.
Based on this plan, China is making every effort to develop the treatment of Alzheimer’s disease, Parkinson’s disease, epilepsy, schizophrenia, depression and other diseases with great social burden. Nowadays, great progress has been made in the study of these diseases.
Education determines the development direction of Science in the future. Brain science is a subject between medicine and engineering. How to cultivate such talents is an important problem facing our country at present. From the current educational situation, all major universities cultivate talents independently, and then cross to brain science for research. Such a development mode is lack of strategy and is difficult to develop to the whole industry.
In order to deal with such problems, Zhejiang University began to implement the “two brain plan” in 2018 to promote the combination of brain science and artificial intelligence. In January 2020, Zhejiang University recently established the school of brain science and brain medicine to solve the most basic education problems.
“Brain science is the most challenging frontier discipline and the fastest growing discipline in the world in recent years.” Duanshumin said that 30000-40000 people attended the annual exchange conference of neuroscience in the United States, which has become the largest discipline. But in China, there are more than 400 undergraduate education majors offered by ordinary colleges and universities, and none of them is a neuroscience major.
“At present, the candidates for Postgraduates of brain science research in China mainly come from undergraduates majoring in biology and biotechnology. They have hardly received knowledge education related to neuroscience at the undergraduate stage, which is not conducive to the development of scientific research.” Duanshumin said. He noted that during the entrance interview of these students, most of them showed an interest in brain science since childhood, but there was no major they could choose at the undergraduate level.
Therefore, it is one of the solutions to the current dilemma to introduce interested students into science and engineering at the undergraduate stage, and then based on clinical education at the master’s stage.
Only by solving the most basic talent problem, can we talk about more in-depth brain science research, and then promote the development of brain science medicine industry. The model of Zhejiang University is undoubtedly a banner, hoping to guide major universities to further develop brain science.
The ultimate goal of both policies and talents is to establish a sound whole industry of brain science and medicine, and to benefit the public with valuable medical technology. In fact, in the past decade, a group of people with lofty ideals have started the exploration of brain science, and have successfully extended their achievements to the fields of neurological diseases, mental diseases, rehabilitation and so on. Among these enterprises, arterial network takes three enterprises, namely boruikang, niunuo technology and Zhentai intelligence, as examples to briefly describe the development of brain science in China.
Boruikang: build a non-invasive brain computer interface system with the highest communication rate
Recognized as a leading enterprise in the industry, boruikang technology is the first company in China to industrialize brain computer interface. Relying on the Neural Engineering Laboratory of Tsinghua University, boruikang has developed a non-invasive brain computer interface system with the highest communication rate in the world, making China occupy a place in the field of international brain computer interface.
In addition to non-invasive, boruikang is also developing a minimally invasive brain closed-loop feedback stimulation system. The system adopts minimally invasive implantation to avoid infection, meets long-term use, and realizes intelligent regulation by closed-loop acquisition and stimulation; Take the treatment of epilepsy as an example, that is, through the characteristic signal warning before epilepsy, trigger electrical stimulation, so as to achieve the effect of inhibiting epilepsy. At the same time, the company is also working on the research and development of brain computer intelligent active rehabilitation system, which will be applicable to the rehabilitation of stroke and other neurological diseases and the intervention treatment of ADHD.
Niunuo Technology: a medical EEG data service provider integrating software and hardware with strong scientific research
With the goal of solving clinical needs and improving patient value, niunuo technology has carried out a two-way software and hardware development mode based on AI algorithm and focusing on medical EEG data services in the product line. Dai explained to the arterial network: “At the beginning, we only worked on algorithms, but we encountered many problems in the process of data collection. First, quality control problems occurred in EEG data collection due to differences in acquisition equipment. Second, many portable scenarios are of great application value, but it is difficult for many manufacturers to quickly follow up this demand and provide technical support for small-scale tests. Third, EEG data needs to be converted through multiple algorithms and eliminated Interference can form an analyzable signal. To grasp the signal of a certain characteristic value, you need to understand the processing of small signals by yourself, which cannot be bypassed by the understanding of the equipment. “
Scenarios and technologies are constantly developing through exploration. In the future, niunuo technology will continue to explore new business paths, deepen cooperation with internet medical, pharmaceutical enterprises, insurance companies, etc., and use AI capabilities to assist EEG services in the implementation of medical systems at all levels. The breadth and depth of indications will not only involve the diagnosis and treatment of traditional brain diseases, such as epilepsy The differential diagnosis of Alzheimer’s disease and other related neurological diseases will also promote the wider popularization and application of medical EEG services in psychiatric specialties, pediatric neurology specialties, severe diseases and emergency departments, and realize the comprehensive benefits of high and new technologies.
Zhentai Intelligence: “bci+vr+ robot” helps patients recover
BCI is just needed in the field of rehabilitation medicine and has already attracted the attention of numerous researchers and enterprises. In the 1990s, scientists began to carry out relevant research; In 2000, EEG based biofeedback products began to be applied in the medical field; In 2014, Giuliano, a paraplegic youth, kicked off for the world cup wearing an exoskeleton machine suit; Today, this industry has entered the product stage.
At present, the company’s brain control intelligent rehabilitation solution is mainly used for the rehabilitation of neurological diseases. In the pre clinical stage of scientific research, there have been more than 1000 trial patients. The research shows that after the rehabilitation treatment of BCI system, the cognitive status, motor ability, balance and muscle tension of patients have been significantly improved. Next, the solution will gradually expand to the whole scene of physical therapy, occupational therapy and other rehabilitation training, and customize the rehabilitation treatment plan for patients accurately through the full cycle of EEG data acquisition and evaluation.
Write at the end
Back to the original question, how far is BCI technology used for disease “treatment”? You already have the answer in your mind.
In general, the breakthrough of neuralink is certainly gratifying, but we should also rationally look at the challenges and limitations of BCI technology in medical applications at this stage. Therefore, for the research in the medical field in China, it may be a more practical way to adhere to the “medical brain” to make a breakthrough, focus on disease diagnosis and treatment, and accumulate disease databases.
After all, the medical field may not be as romantic as musk.
Step by step, this is the rhythm and speed that medical treatment should have.