"Science Everyone" column | How to link the brain and the machine: the future of the magic brain chip is unlimited

by TechnGrandpa 19 days ago in tech

Elon Musk neuralink,a new brain computer interface device

"Science Everyone" column | How to link the brain and the machine: the future of the magic brain chip is unlimited

Last weekend, Musk released Neuralink, a new brain-computer interface device. You only need a small cut on your scalp, and you can implant your brain's Fitbit through a coin-sized device, which can be directly controlled through a mobile phone app. At present, this device has been successfully tested on piglets and has been approved by the FDA for human brain experiments. In the future, this device is likely to not only help people with disabilities control their prostheses, but also to fight against neurological problems such as epilepsy, major depression, autism, Alzheimer's disease, and Parkinson's syndrome.

Many Internet users say that this technology of linking brains and machines is almost a sci-fi film and television work into reality.

In fact, this kind of technology has been successful in animal experiments as early as 2008. No matter what kind of appearance and path, how to convert the "ideas" in the brain into instructions that can be recognized by the machine and realize transmission. A link that cannot be bypassed by technology has always been the key point for scientists to seek breakthroughs.

In fact, not long ago, the live-action version of "Alita" appeared on Weibo hot search. This is Tilly, a young girl from the UK. She had her arms amputated when she was one year old because of sepsis. But 12 years later, she became the first person in the UK to have a bionic arm.

The robotic arm has built-in muscle sensors, so Tilly can easily control the various actions and movement modes of the robotic finger through "thoughts." By using a robotic arm, Tilly can take care of his own life and even become a beauty blogger.

Although the flexibility of the robotic arm needs to be improved, Tilly occasionally has accidents where mascara pokes the eyes while helping friends with makeup, but Tilly still likes cool robotic arms very much.

In fact, the intelligent bionic prosthesis used by Tilly has many successful application cases around the world. With the innovation and development of technology, researchers are also looking for more complete bionic methods, such as in addition to In addition to completing daily actions, there must be higher sensitivity and flexibility, and even touch.

For those patients who have not only lost a part of their limbs, but have severely damaged their nervous system, the control of bionic prostheses through muscles and peripheral nerves is far from enough.

They need technology that can link to the brain to issue instructions, clarify user/patient needs or provide feedback. Such a neural interface needs to have high accuracy, long-term stability, safety and reliability.

History of "Brain Machine" Research In

In 1924, German psychiatrist Hans Berger discovered brain waves. It was discovered that the original human consciousness can be read after being converted into electronic signals.

Since then, brain-computer interface research has sprouted, but it was not until the 1960s and 1970s that these technologies began to take shape.

In 1969, a researcher named Eberhard Fitz connected a neuron in the monkey's brain to the dashboard. When the neuron is triggered, the pointer on the dashboard will rotate.

If the monkey triggers the neuron through a certain way of thinking and rotates the pointer of the dashboard in front of him, he will get a banana-flavored ball. Under the temptation of banana balls, the monkey became good at this game.

This monkey, who learned to control neuron triggering through training, accidentally became the first real brain-computer interface subject. The following year, the Defense Advanced Research Projects Agency (DARPA) began to form a team to study brain-computer interface technology. At the end of the 1990s, brain-computer interface research ushered in a new wave.

There were two BCI international conferences held in 1999 and 2002, and there have been successes in connecting the chip to the human brain through surgery and remotely controlling other equipment through the mainframe.

Case. During the same period, the sci-fi movie "The Matrix" was released, in which there was a plot to transfer knowledge, control equipment and even change reality after inserting a cable in the back of the head. This was no longer a brainstorming fantasy back then, but based on existing research Reasonable imagination.

In fact, in the brain, we already have many achievable technologies, and at the same time, there are many new possibilities that are constantly happening.

In addition to controlling prosthetic limbs, there is more to transmit brain information to the outside, and implantable technology will help us do this.

We can use implantable new technologies to treat some intractable brain-related diseases, these diseases include vision loss, hearing loss, epilepsy, and some of the more terrible diseases we are familiar with, including Parkinson’s disease and depression And so on, there are some diseases that can cause human paralysis.

Researchers hope that through the repair of neurons, these signals can be effectively transmitted and controlled by machines. Researchers hope that this technology can be applied to the brain, where these signals can be received, and by implanting nerve-stimulating chips, they can be connected with spinal neurons to restore the brain and body. Features.

Researchers have carried out such experiments on monkeys, hoping that through sensors, processors, and wearable devices, machines that can promote hearing and vision recovery from the outside can be implanted in the body to promote the recovery of brain function.

The interface between the human brain and the machine is called the human brain-machine interface. New technologies have also been applied in medical practice.

No matter what kind of equipment, including equipment for treating epilepsy, biosensors can be well implanted. And application.

Capturing information in the brain

We know that vision is a very important function. If a patient’s eyes are blind, his retina will not be able to receive information.

If we want to solve this problem through implantation technology, we need to further understand the mechanism of retinal imaging.

Through the exploration of retinal function, through brain imaging, through micro equipment, it may be millimeters or even more accurate. Instruments, implant them into our brains, and then use antennas, small radio frequency units, and other small chips to implant different functions in the brain. These chips can improve our hearing, vision, and promote eyes.

The information from the Ministry is imaged in the brain so that blind patients can see what we see.

There are also many emerging areas of this type of research on the visual part of the brain. The human brain can image what the eye sees, so what is the principle that we can see different colors? It is actually due to the transmission of visual signals.

Researchers can better conduct research through effective functional optimization and the implantation of control units and transceivers on the chip. With visual implantation, we can see how specific images are imaged, and the analysis of these functions will finally perform precise repairs in the brain. In order to complete the above process, we first need to understand how to acquire and transmit different signals in the brain.

In this link, different chips are needed to capture these processes in the brain. Different brain chips have corresponding differences. These chips can perform different functions to record electrical signals and can receive brain electrical signals of different frequencies.

The work of the chip allows us to understand what the problems are in different areas of the brain. Through these "tools" corresponding to the corresponding parts of the brain, they may be only two millimeters of very small devices that can help us in the brain. Restore these functions.

I am very happy to share these functions with you, so that you can better understand how we can transmit external signals to the human brain through the use of these chips.

The application of the brain chip

Today we are discussing the use of very small sensors, these very small electronic units can reach the micron level.

Researchers have conducted different experiments on monkeys. As we all know, monkey brains are somewhat similar to human brains.

By conducting similar experiments on monkeys, we can verify whether these chips can guarantee signal transmission. First of all, it is known that the monkeys are not blind. Researchers will decode it first to understand how the monkey sees the information, then imitate it, and finally simulate such a mechanism. By decoding the monkey brain, it can better help the research on the repair of human brain function. For different brainwave spectrums, the chip can accurately record different brain regions. Let's look at the internal structure of the chip again.

As mentioned earlier, this type of brain chip device is very low energy consumption and does not need any batteries, so the chip device can be directly implanted into the scalp to work. Of course, we still have the latest technology to achieve such a system. For example, the signal can be transmitted several meters, and there are many chips that can be used for effective recording and transmission.

Let's look at a typical brain disorder type disease-epilepsy. About 1% of people in the world may have symptoms of epilepsy. There are many people suffering from the pain of such neurodegenerative diseases, so we also pay attention to the treatment of patients with epilepsy.

The problem is that we have to observe the state of each part of the brain, and then perform effective imaging to observe what the state of the brain is when epilepsy occurs, and how to prevent symptoms from occurring before the onset of epilepsy.

At this time, we need to observe Changes in nerves. If a patient with epilepsy suddenly becomes ill while driving, it may cause a serious safety accident.

Therefore, at this time, if the information captured by the brain chip can be sent 10 minutes or 30 minutes in advance, and the patient can be warned through mobile phones or other methods, it may be possible to avoid accidents and minimize losses and injuries. Based on this, we chose to use VNS and RNS as a solution. After the chip is implanted in the brain, the chip needs to observe the brain at all times. Researchers will perform different stimulations on different brain regions to identify the main areas of epilepsy in patients.

If you want more accurate positioning, patients need to go to the hospital for an MRI. Even if a very precise examination is done through the above methods, the seizures are very urgent. We cannot expect to use MRI diagnosis to ensure that a certain patient will not have the disease at the same time every month.

At this time, researchers need to consider how to use external devices and sensors to better monitor the transmission of EEG signals.

The patients went to the hospital to test the brain waves. During the examination, by carefully observing the EEG, you can understand which part of the brain activity causes epilepsy.

Everyone knows that the onset of epilepsy comes from the brain. If different brain electrical signal changes can be detected on the scalp, it can help doctors and researchers determine and locate which part of the brain caused epilepsy.

Brain chips can meet the needs of individual patients. When they are implanted in the brain, they can better understand the disease of each patient in real time.

In addition, nanotechnology is also used in such research to measure the aggregation of nanoparticles in the brain. Generally speaking, when people are thinking, more obvious neural activity occurs in a certain part of the brain.

If it is a patient's brain in the process of epilepsy, a certain part of the brain may be stimulated to produce a large number of neurotransmitters. At this time, the density and intensity of neurotransmitters accumulated due to neuronal activity can be detected to help us Locate which part of the brain caused epilepsy. Researchers have also done experiments on mice, this is the case we tested. By marking the signal given when the brain is onset with different colors, we show the onset of epilepsy in 10-20 seconds, including the impact on the current signal. These pathogenesis can be monitored by electronic signals.

According to the signals, the frequency of brain electronic signals is increased, and researchers can observe symptoms that occur before epilepsy, including epilepsy.

This chip device is very small in size, about two microns, and consumes only six milliwatts of power. Its startup is also very simple and its sensitivity is very high.

future development of machine learning and brain chips

Even though there are many new technologies now, the treatment of epilepsy is still very difficult.

though there are many new technologies now, the treatment of epilepsy is still very difficult.

Although scientists continue to promote treatment methods and methods, accurate prediction of epileptic seizures is still a difficult problem. This is also the case.

The future development direction of medical technology. From the perspective of machine learning, many classification algorithms have been developed very mature, so we can consider using the accumulated brain activity information data for different types of cases to conduct targeted training on machine learning algorithms. Deep learning can further improve our prediction of epilepsy.

The future idea of ​​our team is to send a warning message to the mobile phone through such monitoring. For stroke patients, such brain monitoring equipment is also applicable, hoping to bring real precautionary benefits to patients.

This is only the current research area of ​​our team, and there are many teams around the world that are studying the transmission of brain electrical signals, or other research directions. Our work is mainly focused on, such as the interpretation, analysis, transmission, and prediction of signal sources. We have also tried to make some applications.

These projects have different classifications, different services, and customized products for different patients. . Other research institutions or universities, such as the University of Toronto, are conducting important research in this field.

As scientists and developers, we hope to have more memory for the predictive processor, but for the brain, we still need to develop a smaller size chip.

We can't implant the notebook in the brain, so we need Reducing the available data storage of implantable processors, memory is one of the more complex problems we will solve in the future.

Nowadays, the chips are getting smaller and smaller, and the energy consumption is getting lower and lower. Using deep learning and machine learning, we can better produce smarter chips, better improve the efficiency of the chip and reduce its energy consumption.

We use algorithm optimization to improve the energy efficiency of hardware processing. At the same time, the progress of materials science has also brought us good news. How to efficiently combine algorithms, circuits and devices will be the goal of researchers in the future.

We hope that through these methods, we can achieve machine learning with lower energy consumption in the future. We also hope that in the future the chip will become smaller and smaller, for example, it can reach the size of millimeters or less, which will bring better patients welfare.

In view of the existing technology, the optimization of energy consumption according to different models is a work that West Lake University is doing, and it is also the latest goal and vision of our team.

It is not a new concept to form machine learning through training to predict the condition. This concept has already appeared in 1975, and many scientists have made very bold attempts and efforts in this regard.

For example, Webos specifically proposed a multilayer perceptron model in the neural network back propagation (BP) algorithm in 1981.

Although the BP algorithm was named as the "reverse model of automatic differentiation" as early as 1970, it has not really been effective until now, and the BP algorithm is still a key factor in the neural network architecture until today.

In 2012, a team also made new attempts. With the slow development of technology, including myself, I have been doing research in this area, and I am still continuing these studies at West Lake University.

From 1975 to the present, there have been a large number of research reports on the prediction of epilepsy by researchers from all over the world. These are all evidence of the vigorous development of such research. We also use miniaturized devices to be implanted in the human body and detect nerve signals through miniaturized sensors. Our chip is very small, can be easily implanted or removed, and easy to replace.

We also have research on bladder incontinence to help solve the suffering of patients. The system we designed involves many equipment, such as biological pumps, including equipment for monitoring urinary tube closure, and so on.

At the same time, we can use mobile phones to monitor signals. I have introduced you to the work we are doing, what we can do, and we are also facing great challenges, because the brain is very complex and its functions are also very diverse. This is a major issue for future research on neurological diseases.

tech
TechnGrandpa
TechnGrandpa
Read next: Understanding the Collective Intelligence of Pro-opinion
TechnGrandpa

See all posts by TechnGrandpa