Recent developments in neuroscience and AI technology inspired by the digital twin brain have opened up new possibilities for solving the mystery of intelligence. Now, a research team led by Professor Jiang Tianzi from the Institute of Automation, Chinese Academy of Sciences, has outlined the key components and features of an innovative platform called “Digital Twin Brain.” The platform promises to bridge the biological and artificial intelligence gap and provide novel solutions for both ends. However, let’s dive deep into the article.
Also, you may like to read The Best ChatGPT Alternatives to Consider in 2024
Similarity between biological intelligence and artificial intelligence
One significant similarity between biological intelligence and artificial intelligence is that both belong to network structures. Since the brain is composed of biological networks, researchers hope to use artificial networks to build corresponding digital models or brain “twins” to input knowledge about biological intelligence into the models. The ultimate goal of this move is to “promote the development of general artificial intelligence and promote precision mental health care.” Undoubtedly, this ambitious goal can only be realized with the joint efforts of scientists from various disciplines around the world.
How will scientists use these twin brains?
Using digital twin brains, researchers can explore the working mechanism of the human brain by simulating/regulating the brain to perform various cognitive tasks in different states. For example, they could model how the brain functions at rest and what problems might occur due to disease or devise new ways to regulate brain activity and guide it out of unhealthy states.
Digital twin brains do have a solid biological basis
Although it sounds like science fiction, the digital twin brain has a solid biological basis. It integrates three core elements: a brain map as a structural scaffold and biological constraint mechanism, a multi-level neural model trained on biological data to simulate brain function, and a series of tools to evaluate and update the current “twin” copy.
Interaction through closed loops
These three core elements will continue to develop and interact through closed loops. Dynamic brain mapping can improve neural models to produce more realistic functional simulations. In the past, researchers have validated “twins” made of these models across a growing range of practical applications, including disease biomarker discovery and drug testing. These applications will deliver continuous feedback, improving the brain map and completing the operational loop.
The complex structure of the biological brain
The biological brain has a complex structure and dynamic system, so it is necessary to establish an extremely detailed brain map, including maps of different scales, multiple modes, and even from other species, to grasp the construction logic of digital twins. By comprehensively collecting relevant maps, researchers can deeply explore all aspects of the brain, as well as the connections and interactions between different regions in the brain, and ultimately solve the mystery of the principles of brain organization.
On the other hand, the brain map also represents a constraint; the neural model must be based on the map to achieve “biological plausibility,” which also brings technical challenges.
Jiang Tianzi’s team believes that the brain network group map will become important in developing digital twin brains. In 2016, researchers from the Institute of Automation of the Chinese Academy of Sciences announced that the macro map included 246 brain regions and was moving toward an “extensive and detailed mapping” of brain structure and connectivity.
Digital Twin Brain simulation platforms
At the same time, given that existing brain simulation platforms often lack anatomical underpinnings, the authors believe it will be crucial to design “a set of open-source, efficient, flexible, user-friendly, and atlas-constrained brain simulation platforms.” The platform must be powerful enough to support multi-scale and multi-modal modeling. Of course, there are still many pending.
In recent years, artificial intelligence and brain science have each made rapid progress, changing human lifestyles and extensively promoting the development of science and technology and the improvement of human civilization. The cutting-edge progress in brain science research has revealed the complex relationship between brain structure and function, and the success of artificial neural networks has highlighted the importance of network architecture. Nevertheless, it is said that there is growing optimism that artificial intelligence will surpass human intelligence shortly, but there is still a massive gap between the two types of intelligence.
On September 5, 2023, Jiang Tianzi’s team from the Brain Network Group Research Center of the Institute of Automation, Chinese Academy of Sciences, published a long perspective paper titled “Digital Twin Brain: a bridge between biological and artificial intelligence” in the Science collaborative journal Intelligent Computing, proposing digital twins Digital Twin Brain (DTB) is a transformative platform that bridges the gap between human intelligence and artificial intelligence.
However, the article indicates that artificial neural networks and brain network groups based on network nodes and connections share basic units. We can transfer knowledge of human intelligence to artificial intelligence by creating a network. Also, the digital twin brain bridges human intelligence and artificial intelligence as a brain network—a bridge between.
Structural basis for the emergence of brain functions
Different from focusing on simulation or task performance, the digital twin brain emphasizes that the brain network group map is the basis of the “twin”; that is, it emphasizes the structural basis for the emergence of brain functions, how to generate tasks from brain structures, and how to develop universal artificial intelligence based on biological rationality. Intelligence or machine intelligence behaves more like human intelligence.
Therefore, the article proposes that the digital twin brain should include three core elements: structural basis, functional emergence, and application value (Figure 1). Among them, the brain network group map provides the multi-scale network organization information of the biological brain for the digital twin brain. It is the basis and essential constraint for establishing the digital twin brain. The article introduces the three core elements of the digital twin brain respectively. Furthermore, the structural basis of the digital twin brain, multi-level computing models, and functional simulations from neurons to the whole brain. In addition, simulation of brain functions, dysfunction, brain disease intervention, etc., research progress in this area.
Finally, the challenges and prospects of future digital twin brain research are discussed, emphasizing the interaction of the three cores of the digital twin brain.