The human digital twin brain in the resting state and in action
We simulate the human brain at the scale of up to 86 billion neurons, i.e., digital twin brain (DTB), which mimics certain aspects of its biological counterpart both in the resting state and in action. A novel routing communication layout between 10,000 GPUs to implement simulations and a hierarchical mesoscale data assimilation method to be capable to achieve more than trillions of parameters from the estimated hyperparameters are developed. The constructed DTB is able to track its resting-state biological counterpart with a very high correlation (0.9). The DTB provides a testbed for various "dry" experiments in neuroscience and medicine and illustrated in two examples: exploring the information flow in our brain and testing deep brain stimulation mechanisms. Finally, we enable the DTB to interact with environments by demonstrating some possible applications in vision and auditory tasks and validate the power of DTB with achieving significant correlation with the experimental counterparts.
PDF Abstract