Researchers have developed a highly detailed digital twin of the brain of a two-year-old child with autism spectrum disorder (ASD), creating a virtual model that combines brain anatomy with real-time neural activity. The new system, called FEDE (high FidElity Digital brain modEl), enabled scientists to identify neural characteristics associated with autism that would not have been visible using conventional modelling techniques. The proof-of-concept study highlights the potential of digital twins to support more personalized research into neurological disorders.
Autism spectrum disorder is one of the most common neurodevelopmental conditions, affecting communication, learning and social interaction. Although advances in neuroimaging have provided increasing insight into brain development, researchers have struggled to combine detailed structural information with the dynamic electrical activity of the brain in a single computational model. FEDE was developed to bridge that gap.
Combining MRI and EEG into a virtual brain
To create the digital twin, researchers combined advanced magnetic resonance imaging (MRI) with electroencephalography (EEG) recordings obtained from a two-year-old child diagnosed with ASD. The MRI data were used to reconstruct the child's brain anatomy, while the EEG recordings captured real-world brain activity. The researchers then integrated these datasets using multiple layers of mathematical modelling. The result was a virtual brain capable of simulating not only the child's brain structure but also its electrical signalling and neural dynamics.
Unlike existing brain models, FEDE incorporates information about myelination, the fatty insulating layer surrounding nerve fibres that influences the speed of neural signal transmission. The model also accounts for the electrical properties of the skull, scalp and other tissues. In total, researchers reconstructed the head using twelve different tissue types, creating a highly detailed three-dimensional representation of how electrical signals travel from deep brain structures to the scalp.
The digital twin accurately reproduced the child's recorded EEG signals, matching both the timing of brain activity and the regions where those signals originated. According to the researchers, this unified approach provides a much more realistic representation of brain function than conventional virtual brain models.
Hidden neural signatures of autism
After calibrating the digital twin to match the child's measured brain activity, the researchers identified several neural characteristics associated with autism. One of the most striking findings was that the simulated brain exhibited background electrical noise approximately one hundred times higher than that of a standard brain model. The researchers also observed a markedly elevated excitation-to-inhibition (EI) ratio, approximately three times higher than normal, indicating an imbalance between neural signals that activate brain circuits and those that suppress them.
Such neural overactivity has previously been associated with autism, but FEDE enabled the researchers to estimate patient-specific abnormalities in synaptic transmission and synaptic plasticity, the mechanisms through which neurons communicate and adapt over time. These subtle irregularities could not be detected directly from conventional imaging or EEG data alone.
According to the research team, the ability to connect detailed anatomy with physiological brain activity within a single model provides new opportunities to study the biological mechanisms underlying autism in individual patients.
Toward personalized brain modelling
Digital brain models are increasingly being explored as tools for precision medicine, offering the possibility of simulating disease processes and eventually supporting diagnosis or treatment planning. However, until now, brain imaging pipelines, electrical head models and neural activity simulators have largely existed as separate technologies.
By integrating these components into one workflow, FEDE demonstrates that it is possible to generate a patient-specific digital twin capable of reproducing both structural and functional characteristics of the brain. The researchers emphasize that the current study represents an initial proof of concept based on a single child with ASD. Larger studies involving more diverse patient populations will be required before the technology can be validated for clinical use.
Nevertheless, the work illustrates how digital twin technology may contribute to a deeper understanding of complex neurological disorders. In the longer term, such models could help researchers identify individual disease mechanisms, improve patient stratification and support the development of more personalized diagnostic and therapeutic approaches for autism and other brain disorders.
Digital twin for preventive care
Last year, researchers at the Weizmann Institute developed an AI-powered digital twin that uses the world’s largest and most comprehensive health database to predict future disease risk and support personalized preventive care. The model is based on more than 260 billion health data points collected over 25 years from over 30,000 volunteers. It integrates information from 17 body systems, including genetic, metabolic, imaging and microbiome data, to simulate individual health trajectories.
One of its first applications is calculating the biological age of specific organs, enabling earlier detection of conditions such as prediabetes, often years before conventional tests. The researchers also identified sex-specific aging patterns, including accelerated biological aging around menopause. Ultimately, the digital twin could help clinicians tailor prevention strategies and evaluate the long-term impact of lifestyle changes or treatments before disease develops.
