Periodic Reporting for period 1 - PINGED (Personalizing virtual brains with neurodegenerative disease: noninvasive stimulation approach)
Okres sprawozdawczy: 2023-10-01 do 2025-09-30
Alzheimer’s disease (AD) is a progressive neurodegenerative disorder that affects millions globally, posing significant challenges to healthcare systems and societies. Despite advances in early biomarkers, current diagnostic tools often fail to capture the diversity of neurophysiological subtypes and the individualized nature of disease progression. Virtual brain twins have the capacity to integrate heterogeneous data and capture the inter-individual variability, but their application in the case of AD is challenging due to non-identifiability of relevant parameters from spontaneous brain activity.
The PINGED project aims to address this gap by developing a personalized virtual brain modeling framework that integrates multimodal neuroimaging data. Specifically, it combines resting-state functional MRI (fMRI) with EEG responses to noninvasive brain stimulation techniques such as temporal interference (TI) and transcranial magnetic stimulation (TMS). The perturbational approach improves the estimation of mechanistic parameters that reflect an individual’s position along the AD progression trajectory by utilizing the information contained in the brain response as compared to spontaneous activity.
By focusing on individual variability and mechanistic modeling, PINGED contributes to the broader goals of personalized medicine and precision diagnostics. The project’s outcomes are expected to support earlier and more accurate stratification of patients, inform clinical decision-making, and ultimately improve therapeutic outcomes. In doing so, PINGED aligns with strategic EU priorities in health innovation and digital transformation.
The PINGED project aims to address this gap by developing a personalized virtual brain modeling framework that integrates multimodal neuroimaging data. Specifically, it combines resting-state functional MRI (fMRI) with EEG responses to noninvasive brain stimulation techniques such as temporal interference (TI) and transcranial magnetic stimulation (TMS). The perturbational approach improves the estimation of mechanistic parameters that reflect an individual’s position along the AD progression trajectory by utilizing the information contained in the brain response as compared to spontaneous activity.
By focusing on individual variability and mechanistic modeling, PINGED contributes to the broader goals of personalized medicine and precision diagnostics. The project’s outcomes are expected to support earlier and more accurate stratification of patients, inform clinical decision-making, and ultimately improve therapeutic outcomes. In doing so, PINGED aligns with strategic EU priorities in health innovation and digital transformation.
During the reporting period, the PINGED project conducted a comparative evaluation of mean field brain models with respect to their ability to reproduce transcranial evoked potentials (TEPs), capture relevant data features, and reflect individual trajectories along the Alzheimer’s disease progression spectrum. This comparison informed the selection of modeling strategies best suited for personalized diagnostics.
To ensure the reliability of model-based inferences, the project systematically assessed parameter identifiability within the proposed framework. This involved generating synthetic datasets and benchmarking the inference pipeline under controlled conditions, enabling robust estimation of subject-specific mechanistic parameters.
Building on these foundations, a multi-modal pipeline was developed that integrates EEG and MRI data with electric field modeling. This pipeline was applied to empirical data acquired from collaborators in Santa Lucia Foundation (Rome, It.), demonstrating its capacity to fit real-world patient data and extract meaningful insights into disease dynamics. The results support the feasibility of personalized virtual brain modeling in clinical settings.
To ensure the reliability of model-based inferences, the project systematically assessed parameter identifiability within the proposed framework. This involved generating synthetic datasets and benchmarking the inference pipeline under controlled conditions, enabling robust estimation of subject-specific mechanistic parameters.
Building on these foundations, a multi-modal pipeline was developed that integrates EEG and MRI data with electric field modeling. This pipeline was applied to empirical data acquired from collaborators in Santa Lucia Foundation (Rome, It.), demonstrating its capacity to fit real-world patient data and extract meaningful insights into disease dynamics. The results support the feasibility of personalized virtual brain modeling in clinical settings.
PINGED advances the state of the art by implementing a personalized modeling pipeline based on the Jansen-Rit neural mass model. This model successfully captures both transcranial evoked potentials (TEPs)—quantified via e.g. Area Under Curve and Hjorth parameters—and resting-state EEG features such as alpha frequency peak, relative alpha power, and functional connectivity. Crucially, the model includes a parameter related to the synaptic strength of pyramidal neuron inputs, which serves as a mechanistic marker along the Alzheimer’s disease (AD) progression trajectory.
The project developed and validated a multi-modal inference pipeline that integrates EEG and MRI data with electric field modeling. Systematic evaluation of parameter identifiability revealed that different model parameters exhibit varying sensitivity across the AD trajectory. This finding underscores the importance of trajectory-aware modeling, as certain parameters become more informative at specific stages of disease progression.
Empirical validation demonstrated that the model can accurately fit real-world data and differentiate between AD and healthy control (HC) group-level responses. These results confirm the feasibility of using personalized virtual brain models for stratification and diagnosis in clinical settings, positioning PINGED as a pioneering effort in digital neurodiagnostics.
The project developed and validated a multi-modal inference pipeline that integrates EEG and MRI data with electric field modeling. Systematic evaluation of parameter identifiability revealed that different model parameters exhibit varying sensitivity across the AD trajectory. This finding underscores the importance of trajectory-aware modeling, as certain parameters become more informative at specific stages of disease progression.
Empirical validation demonstrated that the model can accurately fit real-world data and differentiate between AD and healthy control (HC) group-level responses. These results confirm the feasibility of using personalized virtual brain models for stratification and diagnosis in clinical settings, positioning PINGED as a pioneering effort in digital neurodiagnostics.