Background: Lewy body dementia (LBD) is the second most common neurodegenerative

dementia and remains underrecognized in clinical practice. Although resting-state

electrophysiologic slowing is a promising biomarker in LBD, objective markers of its temporal

instability are limited. Cognitive fluctuations (CF) are a core feature of LBD, yet the

relationship between spectral slowing and CF severity remains poorly understood. We

investigated whether resting-state MEG with hidden Markov modeling (HMM) can identify

state-dynamic and spectral signatures associated with LBD and CF severity.

Methods: We studied 60 participants across four groups: LBD (n=9; 71±6 years; 7 male),

Parkinson’s disease (PD, n=11; 72±6 years; 7 male), Alzheimer’s disease or amnestic mild

cognitive impairment (AD/aMCI, n=8; 72±11 years; 5 male), and healthy controls (HC, n=32;

71±7 years; 11 male). Cognitive fluctuations were quantified with the Clinician Assessment of

Fluctuation (CAF). Each participant underwent five 2-minute resting-state

magnetoencephalography (MEG) runs. Data were preprocessed and source reconstructed with

an LCMV beamformer into 52 cortical parcels. An 8-state HMM was used to identify transient

brain states. Fractional occupancy (FO), switching rate (SR), and transition-probability metrics

were derived. Spectral slowing was quantified using power spectral density and theta/beta

ratio (TBR).

Results: Compared with PD, AD/aMCI, and HC, the LBD group showed a shift toward lower

oscillatory frequencies, with significantly higher TBR (P <0.05). State 4 emerged as the

dominant LBD state, with significantly higher FO and SR than PD and HC (Bonferroni-adjusted

P <0.05); similar increases relative to AD/aMCI did not reach significance (Figure 1). LBD

showed greater deviation of the run-averaged transition-probability matrix than HC,

indicating less stable state transitions (Figure 2). State 4 showed the greatest TBR elevation

in LBD relative to the other groups (Figure 3A). Within LBD, higher CAF scores correlated with

higher TBR in multiple areas, including visual, inferior parietal, left inferior frontal, and right

insular cortices (Figure 3B).

Conclusion: LBD was characterized by a dominant brain state, increased transition instability,

and spectral slowing. Greater CF severity was associated with stronger regional slowing.

These findings support resting-state MEG state dynamics and spectral slowing as candidate

measures of LBD and CF, with utility for di"erential diagnosis and treatment monitoring.