Where are you using WebAssembly?
Wasm promises to let developers build once and run anywhere. Are you using it yet?
At work, for production apps
At work, but not for production apps
I don’t use WebAssembly but expect to when the technology matures
I have no plans to use WebAssembly
No plans and I get mad whenever I see the buzzword

‘Digital Twin Brain’ Could Bridge Artificial and Biological Intelligence

Such brain atlases could help us understand how brain regions are interconnected and how they interact at various levels of granularity.
Nov 23rd, 2023 3:00am by
Featued image for: ‘Digital Twin Brain’ Could Bridge Artificial and Biological Intelligence

A growing number of experts in the field of artificial intelligence are looking to neuroscience for inspiration when it comes to finding new types of algorithms and network architectures. In particular, ideas like artificial neural networks and deep learning take their cues from how human brains function and learn, and may be key to developing human-like intelligence in machines, or what is known as artificial general intelligence (AGI).

Mapping the human brain correctly will likely help in this ongoing endeavor toward developing artificial general intelligence, especially in the creation of a “virtual brain” that can simulate the underlying interactions between neurons.

However, numerous attempts over the years to develop an accurate virtual model of the brain have been hampered by the exclusion of genetic factors and the lack of sufficient cross-integration between models of brain function across various scales and experimental methods.

But these oversights may soon be corrected with a platform for building a “digital twin brain” recently introduced by a research team from the Institute of Automation of the Chinese Academy of Sciences.

The team’s paper, which was recently published in Intelligent Computing, describes how their digital model of the human brain would bridge the gap between artificial and biological intelligence by incorporating recent discoveries in neuroscience, thus helping to build better AI models and more refined neurological and cognitive simulations, as well as advancing the field of precision medicine, where mental health treatments would be personalized according to the individual patient.

“Cutting-edge advancements in neuroscience research have revealed the intricate relationship between brain structure and function, and the success of artificial neural networks has highlighted the importance of network architecture,” wrote the team. “It is now time to bring these together to better understand how intelligence emerges from the multi-scale repositories in the brain. [..] By mathematically modeling brain activity, a systematic repository of the multi-scale brain network architecture would be very useful for pushing the biological boundary of an established model.”

As that systematic repository, the team’s digital twin brain (DTB) would be capable of simulating various states of the human brain in different cognitive tasks at multiple scales, in addition to helping formulate methods for altering the state of a malfunctioning brain.

As the team points out, such a virtual approach would be critical in finding new neurological treatments. “The advantages of this research approach lie in the fact that these methods not only simulate [biologically plausible] dynamic mechanisms of brain diseases at the neuronal scale, at the level of neural populations, and at the brain region level, but also perform virtual surgical treatments that are impossible to perform in vivo owing to experimental or ethical limitations. [..] Collectively, computational models have the potential to provide insights into the mechanisms underlying brain diseases and aid in the design of interventions.”

The Role of the Brain Atlas

In developing the digital twin brain, the team incorporated prior and current research from atlases of the human brain — or neuroanatomical maps of the biological brain. In particular, the research team is basing their digital replica on information gleaned from the Brainnetome Atlas, which was launched back in 2016 under the guidance of the team’s lead researcher, Tianzi Jiang.

According to the team, the main advantage of Brainnetome is that synthesizes various repositories of information taken from various imaging techniques and experimental methodologies, allowing it to better represent brain connectivity at different scales and modalities across the brain’s 246 subregions, including structural and functional patterns between them.

Brain atlases “can help us understand how brain regions are interconnected and how they interact at various levels of granularity,” explained the team. “Such insights are crucial for modeling brain dynamics and simulating complex neural processes.”

The team also noted that basing AI models on information adapted from brain atlases could provide a significant advantage: “Networks trained using biologically realistic connectivity often outperform those trained on random networks. Integrating data from different imaging modalities can provide a comprehensive view of brain structure, connectivity, and activity. This multimodal approach enables us to capture complementary information about the brain, enhancing our ability to model and simulate neural activity with higher accuracy.”

Going forward, the team says that it will be crucial to construct their digital twin brain platform as a closed-loop, yet dynamic, resource that will be user-friendly, adaptable and open source.

Additionally, the team envisions that the DTB will be continuously validated and improved when it is put to the test in real-world applications like biomarker identification and pharmacological testing and made more computationally efficient over time to ensure effective modeling across different scales and modalities.

Group Created with Sketch.
THE NEW STACK UPDATE A newsletter digest of the week’s most important stories & analyses.