Multimodal Brain Aging Models And Their

Biophysical Mechanisms

Sebastian Moguilner1,2, Carlos Coronel-Oliveros1, and Agustin Ibanez1

1Multi-Partner Consortium to Expand Dementia Research in Latin America

2Department of Anesthesia, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA

BACKGROUND

RESULTS

• Brain clocks quantify brain-age gaps (BAGs) (predicted age

minus chronological age calculated from neural features)

• These BAGs can be used to measure accelerated aging,

monitoring deviations from healthy brain trajectories

• Most evidence comes from high-income settings, overlooking

global diversity

• Diversity shapes brain health, yet few studies pair

neuroimaging with biophysical modeling [1]

• Whole-brain models can map gaps to neural mechanisms [2]

• The biophysical mechanisms driving healthy aging deviations

remain poorly understood

• The biophysical model showed that the participants with

higher BAGs had higher firing rates/hyperexcitability (P =

0.00014).

• In patients, we observed that AD is characterized by a global

shift towards lower firing rates/hypoexcitability (P = 0.008)

METHODS

• SVM predicting BAGs achieved Pearson’s r = 0.727, Cohen’s f2

= 1.118, mean absolute error (MAE) = 7.723 and root-mean-

square error (RMSE) = 9.833

• By categorizing countries into low- and high-income groups

based on the median of the income distribution, we observed

accelerated brain aging in low-income countries (5.44 BAG

difference [2.52, 6.56]95% , t(1,183) = 7.40, P = 2.54 × 10−13

, D =

0.45, FDR-corrected)

• When analyzing variables such as age, income, education, and

sex, we observed disparities, with higher BAGs among

participants in the global south

• No BAG differences were found between females and males in

the global north. However, accelerated aging in females was

observed in the global south (−4.78 BAG difference, [−7.49,

−2.07]95% , t(469) = −3.46, P = 5.98 × 10−4

, D = 0.32, FDR-

corrected P values)

• Accelerated brain aging is more pronounced in AD than HCs

(9.12 BAG difference, [6.80, 11.44]95% , t(217) = 7.72, P = 4.16

× 10−13

, D = 1.05, FDR-corrected P values) and in bvFTD (6.88

BAG difference, [4.41, 9.35]95% , t(174) = 5.08, P = 9.34 × 10−7

,

D = 0.83, FDR-corrected P values)

• We used EEG data from participants across ten countries from

the global north and the global south (N = 1,399):

• Healthy controls (HC) (North N = 724, South N = 516)

• Alzheimer’s disease (AD) (South N = 102)

• Behavioral variant frontotemporal dementia (bvFTD)

(South N = 57)

• Countries involved:

• Global North = Turkey, Greece, Italy, the United Kingdom,

and Ireland

• Global South = Cuba, Colombia, Brazil, Argentina, and Chile

• We used support vector machine (SVM) algorithms to compute

BAGs and biophysical semi-empirical modeling to analyze the

neural mechanisms driving those BAGs

• Biophysical modeling:

• Generative whole-brain model fit to each participant

• Simulates the excitatory/inhibitory balance to check for

hyperexcitability and hypoexcitability effects driving BAGs

• Statistical analysis:

• Student t-tests for comparisons between groups and

conditions, Cohen’s d for effect sizes

• Evaluated BAGs associations with diagnosis, country, and sex

Figure 2. Biophysical mechanisms. Generative whole-brain modeling separates

accelerated healthy aging (hyperexcitability) from dementia (hypoexcitability).

CONCLUSIONS

• Brain clocks capture diversity and burden across aging and

dementia

• The results suggest that, in dementia, neurodegeneration is

characterized by global hypoexcitability

REFERENCES

1. 2. Coronel-Oliveros C, Moguilner S, Hernandez H, et al. Diversity-sensitive brain

clocks linked to biophysical mechanisms in aging and dementia. Nat Ment

Health. 2025;3(10):1214-1229. doi:10.1038/s44220-025-00502-7

Moguilner S, et al. Biophysical models applied to dementia patients reveal

links to loss of neural inhibition. Alzheimers Res Ther. 2024;16:79.

ACKNOWLEDGEMENTS

Figure 1. Diversity-sensitive brain-age gaps. Model performance estimating BAGs,

and regional differences with aging, income, education, sex, and groups.

Supported by ReDLat, the Global Brain Health Institute (GBHI), and

NIH/National Institute on Aging.