Alzheimer’s disease (AD) and other neuropsychiatric disorders impose a major global burden, yet current treatments remain limited. Beyond amyloid and tau pathology, AD is increasingly understood as a disorder of large-scale brain networks, marked by impaired gamma oscillations and excitation–inhibition imbalance. These circuit-level alterations affect cognition and represent promising therapeutic targets.
Transcranial alternating current stimulation (tACS), especially at 40 Hz, can modulate these networks, but current approaches do not account for individual variability in anatomy or physiology. Neurotwin addresses this gap by developing personalized hybrid brain models (“neurotwins”) integrating electric-field modelling, neural-mass physiology, and multimodal neuroimaging to predict and optimize stimulation effects.
The project delivered a multiscale modelling framework capable of (i) mapping electric fields with high anatomical precision, (ii) linking stimulation to laminar, mesoscale, and whole-brain dynamics, (iii) assimilating multimodal human and animal data, and (iv) defining quantitative targets representing healthy neural dynamics. This enables individualized, mechanism-informed, home-deployable neuromodulation strategies.
Overall objectives
1. Develop a validated multiscale modelling framework.
2. Build personalization/data-assimilation pipelines.
3. Define objective functions for healthy dynamics.
4. Conduct in-vivo and in-human experiments.
5. Run a clinical pilot of personalized tES at home.
6. Deliver an operational engineering pipeline.
Conclusions of the Action
NEUROTWIN delivered the first full platform for digital-twin–based personalized brain stimulation, combining modelling, experimentation, engineering, and clinical validation.
1. Modelling framework: A complete architecture—laminar NMMs, whole-brain models, and personalization pipelines—was achieved, reproducing AD biomarkers (e.g. oscillatory changes during the course of disease) and generating actionable stimulation targets.
2. Experimental validation: Rodent optogenetics, Neuropixel recordings, and 40 Hz tACS studies established mechanistic grounding. Human EEG, TMS-EEG, and multimodal MRI validated effects across scales and informed personalization.
3. Clinical pilot in AD: The home-based personalized 40 Hz tACS trial (n=30) showed high feasibility, excellent safety, and strong adherence. While clinical endpoints did not differ between conditions, neurophysiological markers (gamma power, cortical reactivity, connectivity) showed stimulation-specific effects consistent with model predictions.
4. Engineering pipeline: A regulatory-ready software suite (NADL and Neurotwin Library) was delivered, enabling construction, personalization, optimization, and export of neurotwins. The Xpress pipeline supported the clinical pilot; the full HBM pipeline is released for research.
5. Neurotwin Database: multimodal datasets were curated, anonymized, BIDS-aligned, and deposited with DOIs for long-term reuse.
6. Ethics, dissemination, exploitation: Ethical guidance was completed; dissemination exceeded targets.
Overall conclusion
NEUROTWIN demonstrated the safety, feasibility, and mechanistic validity of personalized, model-driven brain stimulation in AD. The project leaves a mature scientific and engineering foundation, a rich multimodal database, and a clear translational pathway, positioning Europe at the forefront of precision neuromodulation.
