State sensitivity and five-year longitudinal stability of resting-state EEG biomarker candidates in healthy adults

Resting-state EEG is widely used to investigate individual differences in brain function, as it allows repeated measurements with high temporal resolution and minimal task demands. However, validating EEG metrics as biomarker candidates requires more than demonstrating cross-sectional differences at a single time point. It is necessary to evaluate whether each metric is sensitive to transient state changes, such as eyes-open (EO) and eyes-closed (EC) conditions, and whether individual differences remain stable across long-term follow-up. This study investigated EO/EC state sensitivity and five-year longitudinal stability of posterior alpha relative power, alpha peak frequency, theta/beta ratio, and aperiodic exponent using publicly available resting-state EEG data from healthy adults. State sensitivity was evaluated using within-participant EO/EC differences and Cohen's d, while longitudinal stability was assessed using intraclass correlation coefficients (ICC), baseline-follow-up scatter plots, and Bland-Altman plots. The analysis also examined whether the findings were preserved across reduced-channel configurations. Posterior alpha relative power showed the strongest EO/EC state sensitivity (Cohen's d = 1.553) and high five-year longitudinal stability (ICC = 0.843). Alpha peak frequency showed small state sensitivity (Cohen's d = 0.238) and moderate longitudinal stability (ICC = 0.734). Theta/beta ratio increased under the EO condition and showed small-to-moderate state sensitivity (Cohen's d = -0.342) and good longitudinal stability (ICC = 0.772), although outlier sensitivity analysis indicated that its individual-level stability should be interpreted cautiously. Aperiodic exponent also increased under the EO condition and showed moderate-to-large state sensitivity (Cohen's d = -0.761), but its five-year stability was moderate (ICC = 0.668). These findings indicate that resting-state EEG biomarkers should be evaluated separately for state sensitivity, longitudinal stability, and robustness under reduced-channel conditions. Posterior alpha relative power emerged as the most robust candidate marker in healthy adults, but it should be interpreted as a candidate EEG marker rather than as a validated clinical biomarker.