Modeling Alzheimer’s in the Brain: How Neural Mass Models Reproduce EEG Biomarkers

Alzheimer’s disease is usually associated with memory loss, but beneath the surface, it is a disease of brain networks. Long before widespread cell death occurs, the brain’s rhythms (coordinated patterns of activity we can measure with EEG) start to shift. Understanding those shifts could help us detect the disease earlier and develop more targeted interventions.

A recent study from researchers at Neuroelectrics Barcelona and Universitat Pompeu Fabra takes a big step in this direction. Using a computational brain model, they reproduced the characteristic brainwave changes observed in Alzheimer’s disease (AD), linking them directly to the underlying cellular damage caused by amyloid-beta and tau proteins.

The Two Faces of Alzheimer’s Brain Activity

One of the mysteries of AD is that brain activity changes in a two-phase manner:

• Early stages (preclinical/mild cognitive impairment): the brain becomes hyperactive. EEG recordings show increased power in faster rhythms, such as gamma (30–70 Hz) and sometimes alpha (8–12 Hz) bands.

• Later stages (dementia): activity slows down and weakens. Alpha and gamma power decline, while slower waves (theta, delta) become more prominent.

This shift from “too much activity” to “too little activity” reflects the gradual breakdown of balance between excitation and inhibition in brain circuits.

Alzheimer’s Brain Activity

What’s Behind the Shift?

The study emphasizes two key players:

• Amyloid-beta (Aβ) oligomers: small toxic clumps of protein that appear early in the disease. These target parvalbumin-positive (PV) interneurons, a type of inhibitory cell crucial for keeping brain rhythms stable. PV interneurons act like orchestra conductors: when they fail, the brain’s electrical activity becomes noisy and uncoordinated.

• Hyperphosphorylated tau (hp-τ) protein: which spreads later in the disease. Tau pathology damages pyramidal cells, the main excitatory neurons of the cortex. This leads to hypoactivity, reduced firing, and eventually cell loss.

Together, these two pathologies create the biphasic progression seen in EEG: first an increase, then a collapse of rhythmic activity.

Illustration of Alzheimer's disease progression.

Simulating Alzheimer’s with a Brain Model

To connect these mechanisms with EEG signals, the team used a Laminar Neural Mass Model (LaNMM), a mathematical framework that represents populations of excitatory and inhibitory neurons arranged in a cortical column.

By gradually reducing the inhibitory influence of PV interneurons (to mimic amyloid-beta toxicity), the model produced increased alpha and gamma power, just as seen in early AD. As the inhibition dropped further, oscillations slowed and weakened, reproducing later-stage EEG features.

However, the model alone could not explain the reduced firing rates observed in advanced AD. To fix this, the researchers added tau-driven damage to pyramidal neurons. The combined model (PV + pyramidal dysfunction) successfully reproduced the full trajectory: from hyperexcitability to hypoactivity.

LaNMM

Why This Matters

This study shows that:

• EEG biomarkers (especially changes in alpha and gamma rhythms) can be traced back to specific cellular dysfunctions in Alzheimer’s.

• PV interneuron dysfunction may serve as an early biomarker and therapeutic target, detectable long before massive cell death occurs.

• Computational models like LaNMM provide a mechanistic “bridge” between molecular pathology and clinical signals like EEG/MEG.

For clinicians and researchers, this strengthens the case for using EEG as a low-cost, non-invasive tool to monitor disease progression and evaluate interventions.

Conclusion

Alzheimer’s disease alters the brain’s rhythms long before memory fades. By modeling how amyloid-beta and tau disrupt the delicate balance between inhibitory and excitatory neurons, this study demonstrates how computational neuroscience can shed light on the origins of EEG biomarkers.

The take-home message: changes in alpha and gamma brainwaves are not just signals on a screen, they are fingerprints of the underlying cellular dysfunction driving Alzheimer’s progression. With models like LaNMM, we are getting closer to understanding, detecting, and one day intervening in this devastating disease.

References

Sanchez-Todo, R., Mercadal, B., Lopez-Sola, E., Guasch-Morgades, M., Deco, G., & Ruffini, G. (2025). Fast Interneuron Dysfunction in Laminar Neural Mass Model Reproduces Alzheimer’s Oscillatory Biomarkers (preprint). bioRxiv. https://doi.org/10.1101/2025.03.26.645407