Source-space EEG alpha activity reveals brain age gaps due to neurodegeneration and disparity.

Source-space EEG alpha activity reveals brain age gaps due to neurodegeneration and disparity - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Commun Biol . 2026 May 21;9:740. doi: 10.1038/s42003-026-10205-z Source-space EEG alpha activity reveals brain age gaps due to neurodegeneration and disparity Mónica Otero 1 , Felipe I Carriel-Rubilar 2 , Hernan Hernandez 3 , Jhosmary Cuadros 3 , Jorge G Condado 4 , Agustin Sainz-Ballesteros 3 , Hernando Santamaria-Garcia 5, 6, 7 , Agustina Legaz 3, 8 , Agustina Birba 3 , Sol Fittipaldi 3, 9, 10, 11 , Francisco Lopera 12 , John Ochoa-Gómez 12 , David Aguillon 12 , Alfredis González-Hernández 13 , Jasmin Bonilla-Santos 14 , Rodrigo A Gonzalez-Montealegre 15 , Görsev G Yener 16, 17, 18 , Bahar Güntekin 19, 20 , İlayda Kıyı 21 , Tuba Aktürk 22, 23 , Ebru Yıldırım 24, 25 , Lütfü Hanoğlu 26 , Renato Anghinah 27 , Pedro A Valdes-Sosa 28, 29 , Ronaldo Garcia-Reyes 28, 29 , Javier Escudero 30 , Susanna Lopez 31 , Robert Whelan 9, 11 , Alberto Fernández 32 , Adolfo M García 8, 10, 11, 33 , David Huepe 34 , Marcio Soto-Añari 35 , Eduar Herrera 36 , Daniel Abasolo 37 , Nicolás Rubido 38 , Ruaridh A Clark 39 , Wael El-Deredy 40 , Jesús M Cortes 4 , Mario A Parra 41 , Claudio Babiloni 31, 42 , Agustin Ibanez 3, 8, 10, 11, 19, 43, 44, ✉ , Pavel Prado 45, ✉ Author information Article notes Copyright and License information 1 Facultad de Ingeniería, Universidad San Sebastián, Santiago, Chile 2 Doctorado en Biología Computacional, Universidad San Sebastián, Santiago, Chile 3 Latin American Brain Health Institute (BrainLat), Universidad Adolfo Ibañez, Santiago, Chile 4 Computational Neuroimaging Lab, Biobizkaia Health Research Institute, Barakaldo, Spain 5 Pontificia Universidad Javeriana, PhD Program of Neuroscience, Bogotá, Colombia 6 Center for Brain and Cognition, Intellectus, Bogotá, Colombia 7 Hospital Universitario San Ignacio, Bogotá, Colombia and Universidad Nacional de Colombia, Bogotá, Colombia 8 Cognitive Neuroscience Center, Universidad de San Andrés, Buenos Aires, Argentina 9 School of Psychology, Trinity College Dublin, Dublin, Ireland 10 Global Brain Health Institute (GBHI), University of California San Francisco, San Francisco, USA 11 Global Brain Health Institute (GBHI), Trinity College Dublin, Dublin, Ireland 12 Grupo de Neurociencias de Antioquia (GNA), University of Antioquia, Medellín, Colombia 13 Department of Psychology, Master Program of Clinical Neuropsychology, Universidad Surcolombiana Neiva, Huila, Colombia 14 Universidad Cooperativa de Colombia, Arauca, Colombia 15 AG is with the Laboratorio de Neurocognición y Psicofisología, Universidad Surcolombiana, Neiva, Colombia 16 Dokuz Eylül University, Medical School, Neurology Department, Izmir, Turkey 17 Dokuz Eylül University, Health Sciences Institute, Neuroscience Department, Izmir, Turkey 18 Izmir Biomedicine and Genome Center, Izmir, Turkey 19 Department of Biophysics, School of Medicine, Istanbul Medipol University, Istanbul, Turkey 20 Neuroscience Research Center, Research Institute for Health Sciences and Technologies, SABITA, Istanbul Medipol University, Istanbul, Turkey 21 Dokuz Eylül University, Health Sciences Institute, Neuroscience Department, Izmir, Türkiye 22 TA is with the Neuroscience Research Center, Research Institute for Health Sciences and Technologies (SABITA), Istanbul Medipol University, Istanbul, Turkey 23 Section Brain Stimulation and Cognition, Department of Cognitive Neuroscience, Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, The Netherlands 24 Neuroscience and Neurotechnology Center of Excellence (NÖROM), Gazi University, Ankara, Türkiye 25 Istanbul Medipol University, Istanbul, Turkey 26 Department of Neurology, Istanbul Medipol University, Istanbul, Turkey 27 Reference Center of Behavioural Disturbances and Dementia, School of Medicine, University of Sao Paulo, Sao Paulo, Brazil 28 Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation, University of Electronic Science and Technology of China, Chengdu, China 29 Human Brain Mapping Department, Cuban Neuroscience Center, Habana, Cuba 30 Institute for Imaging, Data and Communications, School of Engineering, University of Edinburgh, Edinburgh, UK 31 Department of Physiology and Pharmacology “V. Erspamer”, Sapienza University of Rome, Rome, Italy 32 Department of Legal Medicine, Psychiatry and Pathology, Complutense University, Madrid, Spain 33 Departamento de Lingüística y Literatura, Facultad de Humanidades, Universidad de Santiago, Santiago, Chile 34 Center for Social and Cognitive Neuroscience (CSCN), School of Psychology, Universidad Adolfo Ibáñez, Penalolen, Santiago, Chile 35 Universidad Católica San Pablo, Arequipa, Perú 36 Departamento de Estudios Psicológicos, Universidad Icesi, Cali, Colombia 37 Centre for Biomedical Engineering, School of Engineering, Faculty of Engineering and Physical Sciences, University of Surrey, Guildford, UK 38 Institute for Complex Systems and Mathematical Biology, University of Aberdeen, Aberdeen, UK 39 Centre for Signal and Image Processing, Department of Electronic and Electrical Engineering, University of Strathclyde, Strathclyde, UK 40 Center of Interdisciplinary Biomedical and Engineering Research for Health, Universidad de Valparaíso, Valparaíso, Chile 41 Department of Psychological Sciences and Health, University of Strathclyde, Glasgow, UK 42 Hospital San Raffaele Cassino, Cassino, Italy 43 Trinity College Dublin, The University of Dublin, Dublin, Ireland 44 Barcelonaβeta Brain Research Center (BBRC), Pasqual Maragall Foundation, Barcelona, Spain 45 Escuela de Fonoaudiología, Facultad de Ciencias de la Rehabilitación y Calidad de Vida, Universidad San Sebastián, Santiago, Chile ✉ Corresponding author. Received 2025 May 8; Accepted 2026 Apr 24; Collection date 2026. © The Author(s) 2026, modified publication 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13226700  PMID: 42162258 Abstract Brain clocks are promising tools for evaluating brain health. However, most current methods rely on structural neuroimaging. Functionally based approaches remain scarce, especially for assessing age-related neurodegenerative diseases. This study examines whether the brain age gap (BAG), the difference between chronological and predicted brain age, reflects neurodegeneration when estimated from electroencephalographic resting-state (rsEEG) α -oscillations, a well-established marker of brain functional aging. It also explores whether α -based brain clocks reflect sociodemographic diversity and structural inequality. The BAG was computed using spectral descriptors of α -activity in the rsEEG source space of 1228 healthy participants, individuals with mild cognitive impairment (MCI), and patients with Alzheimer’s disease or behavioral variant frontotemporal dementia, residing in 10 countries with varying levels of structural inequality. BAGs are increased in MCI and dementia groups, particularly in posterior cortical regions. Structural inequality emerges as the strongest predictor of BAG, surpassing cognition, education, and sex. The findings indicate that an α -oscillation-based brain clock provides a sensitive functional marker of brain aging, capable of capturing neurodegenerative processes as well as the impact of social disparities. This scalable, accessible approach to brain health shows promise for translational use and population-wide screening in underserved, resource-limited settings. Subject terms: Cognitive ageing, Dynamical systems Alpha-oscillation EEG brain clocks reveal accelerated brain aging in dementia, while country-level inequality strongly shapes brain age gaps, supporting scalable screening in underserved, resource-limited settings. Introduction AGING comprises the progressive deterioration of physiological functions vital for survival and adaptation 1 . This functional decline varies across physiological systems, suggesting that biological age can be a more reliable indicator of disease susceptibility and mortality than chronological age 2 . In this context, brain clocks, also known as brain age models, compute the discrepancy between an individual’s physiological and chronological brain ages (the brain age gap, BAG). Positive BAGs indicate accelerated brain aging, a hallmark or a risk factor of neurodegenerative diseases, inducing cognitive decline to dementia 3 – 5 . Furthermore, the BAG is influenced by a complex interplay of factors encompassing sociodemographic diversity and socioeconomic disparity 5 . The BAG has traditionally been estimated using structural neuroimaging 3 , 6 – 9 . Brain-age models based on functional brain data remain comparatively scarce, especially in the context of neurodegeneration 5 , 10 , 11 . Functional information, however, is crucial as disruptions in neural dynamics often precede macroscopic structural damage and behavioral alterations during the aging and preclinical phases of neurodegenerative diseases. Recent developments in brain-age modeling, particularly studies using resting-state magneto- and electroencephalography (rsM/EEG), have provided a methodological framework for this purpose 5 , 12 , 13 . The rsM/EEG captures brain oscillatory electrical activity under relaxed psychophysiological conditions, typically recorded with eyes closed in an unconstrained, task-free mental state. BAG estimates from these oscillations have been obtained using machine-learning/deep-learning frameworks 13 . In multimodal brain clocks, incorporating rsM/EEG features adds to the information provided by magnetic resonance imaging (MRI), which is primarily driven by 8–30 Hz oscillations 14 . In adults, dominant rsEEG activity is observed at 8–12 Hz in posterior (parietal and occipital) scalp locations 15 . This activity, the α -oscillations, exhibits significant age-related alterations, mainly characterized by frequency slowing 16 – 18 and reduced spectral power 19 , 20 . The magnitude of posterior EEG α -activity correlates with cognitive decline in patients with Alzheimer’s disease (AD) 21 , 22 , the most prevalent neurodegenerative disease in aging, and distinguishes these patients from healthy control (HC) individuals 23 . Periodic spectral EEG measures are fundamental for capturing individual differences in brain dynamics, outperforming aperiodic EEG components 24 . Likewise, changes in the EEG periodic (oscillatory) activity mostly explain the high-to-low frequency power shift that characterizes AD 25 . In this context, the EuroLAD-EEG Consortium, a platform for dementia research, recently assembled a large database comprising clinical, sociodemographic, and rsEEG data from individuals residing in countries across the Global South and North 26 , 27 . Initial analyses using this database revealed that variations in rsEEG features, including a reduction in α -oscillation power, were associated with advanced age, poorer cognitive performance, and male sex 24 . Follow-up studies demonstrated that national-level socioeconomic factors, particularly structural inequality, can exert a more significant influence than individual-level characteristics on the complexity, connectivity, and spectral properties of rsEEG, including key descriptors of α -activity 28 . Subsequent investigations on functional cortical networks revealed that structural inequality, environmental pollution, and health disparities significantly increased BAG 5 . Differences in BAG between HC individuals and AD patients, and between males and females, were positively associated with country-levels of structural inequality 5 . Therefore, brain clocks derived from rsEEG α -activity may offer valuable insights into the brain aging trajectories associated with neurodegenerative diseases that induce cognitive decline leading to dementia. This approach is grounded on a brief rsEEG recording, followed by cortical source estimation of well-established quantitative rsEEG descriptors, and the application of artificial intelligence-based algorithms that integrate individual characteristics (e.g., sex and educational attainment as a proxy for cognitive reserve) alongside national-level structural inequality. This study uses the EuroLAD-EEG database 26 to investigate whether the BAG estimated from source-space rsEEG α -activity progressively increases along a continuum from HC participants to individuals with mild cognitive impairment (MCI) and patients with dementia (AD or behavioral variant frontotemporal dementia, bvFTD). The spectral descriptors of rsEEG α -activity and the EEG source generators most relevant to constructing the brain age model were identified. Then, the contribution of demographics, educational attainment, cognition, income inequality, and clinical diagnosis to the BAG was investigated. This study introduces a novel, affordable, and straightforward brain clock based on α -oscillations. We used this easy-to-measure EEG metric to examine whether BAG is sensitive to neurodegeneration, sociodemographic diversity, and socioeconomic disparities. The study offers valuable insights into the aging trajectories associated with neurodegenerative diseases 29 , highlighting the potential of using rsEEG α -activity to develop tools for screening the risk of cognitive deficits in older adults, a cost-effective strategy especially relevant in resource-limited settings. Using a multicenter design 26 , the study explores the generalizability of the results across heterogeneous populations, a critical requirement for robust translational applications. Results Increased BAG in neurodegeneration The rsEEG metrics for brain age modeling were extracted from canonical (fixed) and individualized α -frequency bands. The individualized band was defined using two spectral landmarks: the individual alpha frequency (IAF), identified as the peak frequency with maximum power between 6 and 12 Hz, and the theta–alpha transition frequency (TF), defined as the frequency with minimum power within the 1–6 Hz range 30 – 32 . The brain age model for EEG α -oscillations yielded adequate goodness of fit ( F = 75.85, p < 0.001 , R 2 = 0.21, MAE = 6.22, RMSE = 7.78, Cohen f 2 = 0.13). The intercept ( y 0 ) of the regression line between chronological age and brain age increased from HC to MCI to bvFTD (Fig. 1A ), suggesting that individuals with MCI and bvFTD had greater brain age than age-matched HC participants. The regression line for the AD group had a higher slope ( m ) and a lower y 0 than that of the HC group (Fig. 1A ), indicating that AD patients exhibited elevated brain ages compared to age-matched HC peers, particularly at older chronological ages. The clinical diagnosis exerted a statistically significant effect on BAG ( F (3,1257) = 61.16, p < 0.001, Cohen´s f = 0.111) (Fig. 1B ). Pairwise multiple comparisons using the Fisher LSD method ( p < 0.05) revealed that BAG gradually progressed from HC participants to AD patients (Supplementary Tables 1 and 2 ). Fig. 1. Brain age model of source-space EEG α -frequency oscillations. Open in a new tab A Linear regression between chronological age and brain age derived from α-oscillations, shown for each participant group. Individual data points (circles) and the corresponding regression line are depicted, with regression parameters detailed for each group. B BAG obtained in each group of participants. Circles represent individual values. The horizontal bar represents the median and the 25th/75th percentile of the distribution. C Feature importance, ranked by SHAP values. Bar lengths indicate the relative contribution of each feature, with dark and light tones differentiating hemispheric origins. D Topographical map of the brain clock of EEG α -frequency oscillations. Representative regions contributing to the brain age model are highlighted in various brain views. Maps in A were generated using MapChart (mapchart.net) and are used under the CC BY-SA 4.0 license. The icons representing filtering, resampling, and artifact correction in C were obtained from iStockphoto.com through a paid subscription. The human silhouette used in the acquisition schematic of C was obtained from Shutterstock.com through a paid subscription. The source-localization schematic in C was created by the authors to match the silhouette in the acquisition panel, and the brain graphic in the source-localization panel was adapted from Servier Medical Art (SMART) under the CC BY 4.0 license. Some of these licensed third-party elements were also reused in the Graphical Abstract, and the same credit and licensing information applies to those elements. Relevant features for brain clocks of α -oscillations The power of the canonical α 2 (10.5–12 Hz) and the individual low- α (TF to IAF) and high- α frequency sub-bands (IAF to IAF + 2 Hz) 33 dominated the set of features most relevant for the brain clock of rsEEG α -activity (Fig. 1C ). The set of relevant metrics also included the IAF and the TF, the other parameters describing the individual EEG α -frequency band (Supplementary Tables 3 – 6 ). To topographically map the brain clock of α -oscillations, SHapley Additive exPlanations (SHAP) 34 values were computed to quantify the relative importance of each feature within the regression model. Furthermore, anatomically and functionally linked brain compartments were merged into consolidated regions of interest (ROIs) (Supplementary Table 7 ) 24 , 28 . For each ROI, the absolute SHAP values of all contributing features were summed and then rescaled to a 0–1 range, yielding a normalized measure of regional importance in brain age prediction. The occipital, parietal, cingulate, and hippocampal brain regions emerged as the most relevant ROIs of the brain clock of α -oscillations in both HC and clinical groups (i.e., MCI, bvFTD, and AD) (Fig. 1D ). Other contributing regions were the operculum, orbital cortex, and inferior frontal gyrus (Supplementary Table 8 ). Modulating factors of the brain clocks of α -oscillations The country-level structural inequality was quantified using the Gini coefficient 35 for countries where participants resided during EEG acquisition. Three levels of structural inequality were defined: high, medium, and low (Supplementary Table 9 ). Country-level structural inequality had a statistically significant effect on the BAG of the HC group ( F (2,680) = 12.52, p < 0.001, Cohen´s f = 0.185) (Supplementary Table 10 ). The pairwise multiple comparison (Fisher LSD Method, p < 0.05) showed that the mean BAG of HC individuals increased progressively with increasing levels of structural inequality (Fig. 2A , Supplementary Table 10 ). Data quality analyses indicate that the results were not affected by the study’s multicenter design. Structural inequality was not correlated with the Gini coefficient (Supplementary Fig. 1 , Supplementary Table 11 ), and the mean number of EEG acquisition electrodes did not significantly differ across the levels of structural inequality (Supplementary Fig. 2 , Supplementary Table 12 ). The number of EEG acquisition electrodes did not significantly influence the Overall Data Quality (ODQ) of the EEG estimated in the source space (Supplementary Table 13 ), with the ODQ being a metric that combines parameters informing about the maximum amplitude, the presence of segments with constant amplitudes, as well as the standard deviation, correlation coefficient, and signal-to-noise ratio of the signals 36 . Neither ODQ, the number of participants, nor the number of channels predicted Gini (Supplementary Table 14 ). Fig. 2. Group-level statistics of the brain age gap (BAG) estimated from source-space rsEEG α -activity and classification of accelerated vs. delayed brain aging. Open in a new tab A Mean BAG comparison of HC across groups of countries with diverse structural inequality. B Mean BAG variations as a function of the group of participants (diagnosis) and structural inequality. The analysis only considered two categories of country-level structural inequality and two groups of participants (HC and AD). In A and B , the circles represent individual values, and the horizontal bar represents the median and the 25th/75th percentile of the distribution. C Classification of accelerated vs. delayed brain aging in HC. D Classification of accelerated vs. delayed brain aging in the entire sample. In C and D , the panels from left to right illustrate the ROC curve, the confusion matrix, and the relative importance of the predictive variables in the classification. Measures of the classification performance are outlined. acc: accuracy, pred: precision, rec: recall. The box plots represent the mean (solid vertical line) and the 25th/75th percentile of the distribution, with points representing the 5th/95th percentile of the distribution. When the group-level analysis included both HC participants and AD patients, and the categories of country-level structural inequality were restricted to middle and high (Fig. 2B ), both structural inequality ( F (1,823) = 15.71, p < 0.001, Cohen´s f = 0.209) and clinical diagnosis ( F (1,823) = 36.05, p < 0.001, Cohen´s f = 0.138) showed statistically significant effects on BAG (Supplementary Table 10 ). The interaction between these factors ( F (1,823) = 0.08, p = 0.769, Cohen´s f = 0.010) showed no statistical effect. A higher BAG was linked to living in countries with increased structural inequality (Fisher LSD method, p < 0.05). In countries with similar Gini coefficients, the mean BAG of the AD group was statistically significantly higher than that of the HC participants (Fig. 2B , Supplementary Table 15 ). Furthermore, country-level structural inequality, cognitive status (assessed using the raw total score from the Mini-Mental State Examination, MMSE) 37 , educational attainment, and sex contributed to distinguishing between HC individuals with increased vs. delayed brain aging, defined by BAG values above and below the median of the BAG distribution (Fig. 2C ). The classification yielded moderate performance (AUC of ROC curve: 0.76, accuracy: 0.74, precision: 0.72, F1-score: 0.77, recall: 0.84). The country-level structural inequality was the most influential predictor of BAG, followed by the cognitive status, educational attainment and sex (Fig. 2C ). Expanding the analysis to include both HC and clinical groups and incorporating clinical diagnosis as an additional predictive variable also resulted in an adequate classification of individuals with increased vs. delayed brain aging (AUC of the ROC curve: 0.70, accuracy: 0.71, precision: 0.71, F1-score: 0.81, recall: 0.94) (Fig. 2D ). Structural inequality remained the most influential predictor, followed by cognitive status, educational attainment, clinical diagnosis, and sex (Fig. 2D ). Discussion We investigated brain clocks of EEG α -activity in a cohort of 1228 participants (50–90 years) of both sexes with different clinical statuses (i.e., HC, MCI, bvFTD, and AD) and varying cognitive reserve, as reflected by their educational attainment. They were enrolled in 10 countries with different structural inequalities. Results showed that the BAG was significantly related to rsEEG α -activity estimated in the occipital, parietal, cingulate, frontal, and hippocampal brain regions. BAG was also effectively related to diversity and disparity. At the group level, increased BAG was associated with higher country-level structural inequality and neurodegeneration. Furthermore, country-level structural inequality emerged as the most influential predictor of BAG at the individual level, surpassing cognitive status, educational attainment, and clinical diagnosis. This study demonstrates that BAG, estimated using an easily interpretable and clinically accessible EEG signature of brain function, is sensitive to neurodegeneration, individual diversity, and socioeconomic disparities. The model offers insights into age-related changes through well-described oscillatory dynamics, thereby complementing traditional brain age models based on T1-weighted morphometry or diffusion-based connectivity estimates. The α -oscillations measures are well-recognized indicators of aging and cognitive decline. However, this is the first time these measures have been used to develop sensitive brain clocks. The study’s multicenter design enhances the generalizability of the findings across diverse populations, underscoring the importance of socioeconomic variables in assessing brain aging. Group differences in BAG The brain clock of α -oscillations showed adequate goodness of fit (Fig. 1 ), with an MAE within the range reported in previous EEG-based brain age models (5–12 years) 5 , 13 , 38 , 39 . Although higher than the MAE typically observed in MRI-based approaches, this is expected, as functional brain clocks capture dynamic neural processes that are not accessible through structural imaging 4 . Notably, while tightly fitting brain-age models minimize error, they often yield predictions that provide limited information beyond chronological age 6 . Moderate-fitting models, such as the α -frequency brain clock, can exhibit greater sensitivity to neurodegeneration and sociodemographic influences. Furthermore, the MAE reported in this study is consistent with single-site EEG studies 13 despite the added variability of a multicenter design. Multicentric studies enhance the generalizability of results 4 and enable assessment of sociodemographic influences on brain age, thereby providing greater translational value for clinical and population-based applications. The linear relationship between the brain age estimated from α -frequency oscillations in the EEG-source space (Fig. 1A ) and chronological age supports previous studies demonstrating linear associations between age in older adults and the magnitude of those oscillations 19 , a result with potential clinical applications for population-based aging screening. Furthermore, the findings indicate that alpha-band EEG activity reflects meaningful deviations from normative brain aging trajectories among individuals with MCI and dementia, highlighting the role of clinical status in shaping these trajectories. This result aligns with and complements previous studies using other neuroimaging approaches 5 , 40 , 41 . However, the mean BAG differences among groups were modest, with BAG distributions showing substantial overlap (Fig. 2B ). This is not unexpected, as the present brain age model was restricted to a single EEG frequency band. Greater discriminative power can be anticipated from integrative approaches that combine a larger number of spectral descriptors with complementary EEG domains, including complexity and connectivity measures. The variability observed in our BAGs (2.1–2.8 years, Supplementary Table 1 ) aligns with previous reports from both EEG- and MRI-based studies 5 , 13 , where variability depends on the modality and analytic pipeline employed 4 , 38 , 42 . The effect sizes obtained in this study further indicate that factors beyond clinical diagnosis contribute substantially to the explained variance of BAG. This interpretation is supported by our classification analyses, which demonstrated that structural disparity, cognition, and education exert stronger modulatory influences on BAG than clinical diagnosis of brain health (Fig. 2D ). The increased BAG of MCI individuals and patients with dementia can, in part, be attributed to the atrophy of the posterior cortical generators of rsEEG α -frequency oscillations. Abnormalities in those oscillations have been linked to reduced cortical gray matter volume in individuals with MCI and patients with AD 43 . Increased BAG can also reflect disrupted cholinergic activity in the basal forebrain and its connections to the thalamus and posterior cerebral cortex that regulate cortical arousal and vigilance, a hallmark of neurodegenerative conditions such as AD 32 , 44 , 45 . This dysfunction may lead to altered oscillatory dynamics within thalamic and cortical networks that generate the brain α -rhythm 45 , 46 , which critically depend on the integrity of the forebrain cholinergic pathways and their cortical targets 47 . The rsEEG α -predictors of BAG EEG metrics derived from individual α -frequency bands were relevant to brain age estimates (Fig. 1C , Supplementary Tables 3 – 6 ). This is consistent with previous research showing that power descriptors of individual EEG α -activity enable the accurate classification of HC and AD individuals 32 , 48 . Compared with the canonical approach, individualized α -frequency bands more sensitively capture age-related slowing of rsEEG rhythms and the decline in IAF observed in AD and related neurodegenerative disorders 48 , effects driven primarily by changes in periodic oscillatory activity rather than by non-oscillatory (aperiodic) components 25 . Overall, the findings underscore the core role of the low α -frequency sub-band in BAG estimations (Fig. 1C , Supplementary Tables 3 – 6 ). The power of this EEG frequency sub-band has been considered a neural correlate of the participant’s global attentional readiness and aging 19 , 49 . Its reduction may reflect the desynchronization of dominant posterior rsEEG rhythms, a phenomenon associated with neurodegenerative processes, including the progression of AD 50 . Topographic representation of the α -brain clock Key regions governing the brain clock for EEG α -activity, including the occipital, parietal, and cingulate cortices, exhibit decreased and disrupted α -oscillations in aging and neurodegeneration 19 , 51 . In AD patients, the magnitude of α -oscillations has been associated with occipital cortical volume 52 , 53 . Altered EEG α -power has also been linked to hippocampal and parietal atrophy 54 , 55 , both of which are key brain regions for the brain clock of α -oscillations (Fig. 1D ). These results align with previous findings indicating a systematic decline in EEG α -power estimated in the limbic and parietal brain regions as individuals age 19 , a pattern pronounced in AD patients 51 . Moreover, these findings are consistent with computational modeling studies suggesting that EEG α -power and the IAF are contingent upon the number and synchrony of neurons within the cortical generators of α -oscillations 56 . Although the cingulate cortex is cytoarchitectonically and functionally heterogeneous, it exhibits significant atrophy in Alzheimer’s disease when considered as a whole, highlighting its vulnerability to neurodegeneration 57 . The volume of posterior cingulate cortex and other default mode network (DMN) regions positively correlates with rsEEG α -source activity and global cognitive status in older individuals 58 . Compared with HC participants, AD patients exhibit reduced DMN gray matter volume, diminished rsEEG α -source activity in those regions, and lower global cognitive status 58 . Similarly, parietal cortical areas integral to the DMN, including the postcentral and supramarginal gyri, exhibit severe atrophy in patients with bvFTD 59 , 60 . Demographic diversity and structural inequality are reflected in the α -brain clock The BAG estimated from rsEEG α -activity progressively increased with rising levels of structural inequality (Fig. 2A, B ). This finding complements studies suggesting that country-level structural inequality modulates EEG source-space dynamics 28 and that differences in BAG between HC individuals and AD patients widen as country-level structural inequality increases 5 . Higher country-level structural inequality is associated with reduced source-space rsEEG α -power, particularly in the cingulate cortex and the parietal lobe 28 , two of the brain regions with the highest preponderance in the brain clock of rsEEG α -activity (Fig. 1D ). The rsEEG α -activity estimated in regions critical for mapping brain aging trajectories (occipital and parietal lobes) (Fig. 1D ) is also modulated by sex, educational attainment as a proxy of cognitive reserve, and cognitive status 24 . Notably, EEG spectral metrics most affected by cognitive status (power in the α 2 and low- α EEG frequency bands) 24 were also ranked among the most significant contributors to the brain age model (Fig. 1C ). The heterogeneous influence of diversity and disparity on the BAG of HC participants (Fig. 2C ) emphasizes the profound impact of country-level structural inequality on brain dynamics, which can exceed the significance of general cognitive status and educational attainment 28 , 61 . The relatively lower influence of sex on brain age compared with other variables (Fig. 2C ) aligns with findings from other multifactorial analyses conducted across diverse populations 24 , 28 . While BAG varied significantly across participant groups (Fig. 1B ), clinical diagnosis was not the primary determinant of BAG outcomes (Fig. 2D ). Instead, its effect was surpassed by the influence of country-level structural inequality, general cognitive status, and education attainment (Fig. 2D ). Although challenging, this result does not represent an isolated observation since previous studies illustrate that clinical diagnosis can be outweighed by the cognitive status and educational attainment even when brain health is evaluated behaviorally 62 . This emphasizes the complex interplay between physiological and environmental factors that underlie brain aging trajectories and the relevance of adopting multidimensional approaches to assess brain health 63 . Methodological remarks The following methodological limitations should be considered when interpreting the results. The study is characterized by an unbalanced sample, with the recruitment of participants with neurodegenerative diseases confined to specific countries with middle and higher structural inequality. Future research must strengthen cross-country and cross-regional analyses by employing representative cohorts that span broader geographic and demographic contexts. Furthermore, longitudinal studies are imperative for elucidating brain aging trajectories and disease progression 64 . Notably, future studies should incorporate the age of disease onset and disease duration at the time of EEG recording as predictors of BAG, considering that different cognitive impairments are associated with early- and late-onset AD 65 . Similarly, incorporating individual-level demographics and socioeconomic indicators, such as gender identity, ethnicity, household income, and quality of education, will enhance the understanding of social diversity and provide a more comprehensive perspective on disparity beyond country-level indices. Finally, expanding the set of cognitive assessment tools beyond the MMNE to include cognitive social variables such as empathy and compassion will help to capture the full spectrum of mental abilities. EEG acquisition and processing challenges that may affect model outcomes include the use of diverse electrode configurations, the use of an average brain template for EEG source space estimation, and the choice of ROI definitions. EEG scalp topographies derived from different electrode layouts were integrated using a head mesh model of 1082 electrodes 66 , 67 . This is an alternative to spatial harmonization procedures implemented in other multicenter studies, which are based on assigning equivalent electrodes according to distances 68 , 69 . In our study, the number of EEG electrodes was not correlated with the Gini coefficient (Supplementary Fig. 1 , Supplementary Table 11 ) or with the mean Gini index of each of the three country-level structural inequality categories used in BAG comparisons (Supplementary Fig. 2 , Supplementary Table 12 ). As in previous studies using this dataset 24 , 28 , the number of channels used for EEG acquisition did not predict the signal quality in the EEG source space (Supplementary Table 13 ). This is in accordance with analyses conducted in the group of HC, which demonstrate that neither the ODQ, the number of HC individuals per recruitment center, nor the number of EEG acquisition channels predicted the country-level structural inequality (Gini coefficient) when this outcome was assessed using logistic regressions 28 (Supplementary Table 14 ). Future research should leverage high-density electrode arrays and individual MRI data to increase the resolution and robustness of EEG source-space estimates, refining the interpretation of individual differences across diverse populations and advancing our understanding of brain health. Methods Participants The dataset used in this study 26 comprises rsEEG recordings from 1319 individuals residing in Argentina, Brazil, Chile, Colombia, Cuba, Greece, Italy, Ireland, Turkey, and the United Kingdom (Fig. 3A ). A portion of this dataset has been used in studies on dementia 70 – 75 and is publicly available 27 . The Institutional Ethics Committees of each participating center approved the collection and use of de-identified data, with participants providing informed consent in accordance with the Declaration of Helsinki. All ethical regulations relevant to human research participants were followed. Fig. 3. Methodological framework for brain age gap (BAG) estimation using EEG α -oscillations. Open in a new tab A Geographic and group-level characterization of the study sample. Countries were grouped into lower, middle, and higher structural inequality categories based on the GINI coefficient, with distinct colors representing each group. The participant numbers per group are indicated. B Age distribution stratified by group and sex. C Key processing and harmonization steps for the estimation of α -oscillations in the EEG source space. D Definitions of the EEG α -band adopted in the study. The bandwidth limits are presented in each case. E Workflow for constructing the brain age model of α -oscillations, including data splitting and repetition procedures used to obtain brain age estimates for every participant. F BAG computation. The panel illustrates group-level brain aging, defines accelerated aging, and presents an analysis of feature importance alongside a topographical representation of the brain clock for α -oscillations. G Group-level statistics and classification of BAG. Group-level variations in BAG were investigated as a function of the country-level structural inequality and the participant group. Accelerated vs. slowed brain aging was classified using predictive variables such as country- and individual-level measures of diversity and disparity, as well as diagnosis. Participants were categorized as HC, MCI, bvFTD, or AD, and grouped by the Gini coefficient of their respective nations at the time of data collection 5 , 28 , 61 . This coefficient quantifies deviations in income distributions within an economy from perfect equality 35 and ranges from 0 to 100, where 0 indicates perfect country-level equality, and 100 represents complete country-level inequality 76 . Three levels of structural inequality were defined: high, medium, and low (Fig. 3B , Supplementary Table 9 ). This classification, along with information on age, educational attainment, cognitive status, and clinical diagnosis of the participants (Fig. 3C , Table 1 ), was incorporated into subsequent analyses. Table 1. Details of the study sample Group Relative GINI Country N F:M ratio Age Years of education Cognition HC Low Ireland 85 41:44 67.6 ± 5.5 15.2 ± 4.0 29.0 ± 1.0 UK 50 25:25 68.7 ± 7.8 16.1 ± 6.0 – Greece 24 10:14 68.8 ± 5.3 – 30.0 ± 0.0 Italy 22 13:9 61.7 ± 7.9 15.3 ± 5.0 – Middle Cuba 9 7:2 58.3 ± 4.4 12.1 ± 1.8 29.8 ± 0.5 Argentina 40 29:11 66.5 ± 8.3 17.7 ± 2.3 28.9 ± 0.8 Turkey 252 152:100 66.1 ± 8.2 11.0 ± 5.3 29.0 ± 1.2 Chile 50 35:15 70.1 ± 7.1 14.8 ± 4.3 28.8 ± 1.4 High Colombia 77 60:17 59.5 ± 6.4 13.7 ± 5.1 27.9 ± 1.7 Brazil 72 42:30 67.8 ± 10.0 12.1 ± 3.6 27.8 ± 1.5 MCI Middle Turkey 95 45:40 73.2 ± 6.5 8.9 ± 4.4 25.0 ± 3.6 High Colombia 133 111:22 62.5 ± 6.2 8.1 ± 5.4 27.0 ± 1.7 bvFTD Middle Argentina 33 13:20 69.5 ± 9.1 14.8 ± 4.6 26.9 ± 3.8 Chile 18 12:6 70.7 ± 5.9 14.6 ± 3.7 23.1 ± 4.7 High Colombia 6 2:4 66.5 ± 5.8 7.8 ± 6.7 21.8 ± 4.7 AD Middle Argentina 32 19:13 73.8 ± 7.5 9.9 ± 5.0 23.7 ± 3.3 Turkey 96 62:34 74.2 ± 5.1 8.8 ± 4.4 19.9 ± 4.8 Chile 60 26:34 71.2 ± 7.0 12.4 ± 4.9 21.5 ± 4.0 High Colombia 44 30:14 72.7 ± 5.6 13.5 ± 3.8 21.3 ± 3.4 Brazil 72 49:23 80.7 ± 6.6 10.1 ± 4.3 21.0 ± 4.5 Open in a new tab HC healthy controls, MCI mild cognitive impairment, bvFTD behavioral variant frontotemporal dementia, AD Alzheimer’s disease The cognitive status was assessed using the raw total score from the Mini-Mental State Examination (MMSE) 37 , a valuable instrument for general cognitive assessment in clinical and research contexts 77 . The MMSE scores range from 0 to 30, with scores ≤ 24 indicative of abnormal cognitive function. Participants with MCI met the Petersen criteria, with MMSE scores below the cut-off threshold 24 78 . Clinical diagnosis of AD was based on the current criteria of the National Institute of Neurological Disorders and Stroke–Alzheimer’s Disease and Related Disorders (NINCDS-ADRDA) working group for probable AD 79 . Likewise, patients with bvFTD fulfilled the revised criteria for probable bvFTD 80 . EEG recording, processing, and normalization The rsEEG was recorded using scalp electrodes under eyes-closed and general psychophysical relaxation conditions, with participants seated comfortably in an armchair. Electrooculographic activity was also recorded to control eye movements and blinking. Antialiasing, analog filtering, and frequency sampling were performed before digital storage. The EEG acquisition setting used at each recruitment center is reported in Supplementary Table 16 . The rsEEG was processed using an automatic pipeline that included protocols to mitigate cross-site methodological variations 81 – 83 , which has been adopted in studies on brain age 5 , social diversity 24 , country-level socioeconomic disparity 28 , and dementia 5 , 73 , 75 . The EEG was filtered between 0.5 and 40 Hz, re-sampled to 512 Hz, and re-referenced using the reference electrode standardization technique 84 (Fig. 3C ). Blinking, ocular artifacts, and myogenic activity were corrected using ICLabel (a tool for the classification of EEG-independent components into signals and different categories of noise) 85 and EyeCatch (a tool for identifying eye-related scalp EEG maps) 86 . Noisy channels were replaced using weighted spherical interpolations 66 . Z -score transformations of the EEG time series were computed independently for each recruitment center to reduce inter-site variability 70 , 73 . The artifact-free EEG activity derived from the electrode montage used at each recruitment center was interpolated to produce datasets with a 128-radial electrode layout, using a mesh-head model of 1082 virtual electrodes, as implemented in the EEGLAB headplot function 87 . This approach has been employed as a spatial normalization method in studies examining between-subject and between-session variances across EEG acquisition 87 , and has been implemented in previous multicenter EEG studies 5 , 24 , 28 . Transforming low-density electrode configurations into high-density montages implies generating virtual channels that carry redundant information from the original signals. While this transformation does not imply a gain of information, it reduces the dependence of EEG source estimation on the number of electrodes when applying inverse solution models, whose spatial resolution is inherently constrained under low-density electrode setups. EEG source estimation EEG brain sources (Fig. 1C ) were estimated using the standardized low-resolution electromagnetic tomography (sLORETA) method implemented in the LORETA KEY software 88 , 89 . Standardized current density maps were computed within the 6242 voxels (5 × 5 × 5 mm each) of the sLORETA brain model, for each of the 153,600 voltage distributions corresponding to the first five minutes of artifact-free rsEEG. The sLORETA head model consisted of three concentric spheres, and a signal-to-noise ratio of 1 was used as a regularization parameter of the sLORETA transformation matrix. A brain segmentation of 82 compartments was implemented using the Automated Anatomical Labeling (AAL) atlas 90 . The inverse solution for each scalp voltage distribution yielded the estimated current-density time series across all voxels of the sLORETA head model. The time series for voxels corresponding to the same AAL brain region were averaged to yield the mean brain source time series for that specific AAL region 71 , 72 . EEG parameters The normalized power spectral density (nPSD) for each AAL brain region was calculated using Welch’s method, employing a 4-s Hanning window with 50% overlap, resulting in a spectral resolution of 0.25 Hz 91 , 92 . The canonical α -frequency sub-bands were alpha1 ( α 1), ranging from 8.5 to 10 Hz, and alpha2 ( α 2), spanning 10.5–12.0 Hz 33 . The individual α -frequency sub-bands were defined as low- α (TF to IAF) and high- α (IAF to IAF + 2 Hz) 33 (Fig. 1D ). Using the canonical definition of α -frequency sub-bands ensured cross-study comparability in brain aging. At the same time, the individualized approach accounted for the typically lower IAF observed in AD patients compared with HC individuals 30 , 31 . The relative power density (RPD, the mean nPSD) was calculated for each canonical and individual α -frequency sub-band. Additionally, the equivalent percent power (EPP) of each α -frequency sub-band, representing the proportion of nPSD within the sub-band relative to the total nPSD, was computed 33 , 93 , 94 . These parameters (RDP and EPP), IAF, and TF were used for BAG assessments. Reduction of the EEG analytical space Resting state α -activity can be detected in most regions of the human brain, although it typically co-exists with oscillatory activity in other E/MEG frequency bands 95 . To streamline the analysis of α -activity, anatomically and functionally linked AAL compartments were merged into consolidated regions of interest (ROIs), as described in previous studies using the EuroLAD-EEG dataset 24 , 28 . For example, neighboring brain regions with established functional coupling, such as the hippocampus, para-hippocampal area, and amygdala, were grouped, as were areas within a given cerebral lobe, such as the calcarine fissure and surrounding cortex, the cuneus, the lingual gyrus, and the superior, middle, and inferior occipital gyri. This procedure resulted in ten cohesive ROIs (Supplementary Table 6 ), allowing reliable detection of both the IAF and the TF within each ROI. The same ROI selection has been applied in previous studies investigating associations among the EEG spectrum, sociodemographic diversity, and country-level structural inequalities 24 , 28 . Focusing on ROIs that encompass structurally and functionally interconnected areas accounts for the inherently low spatial resolution of EEG inverse solution methods, thereby enhancing data interpretation and facilitating the identification of meaningful patterns in brain activity 24 , 28 . This reduction of the EEG analytical space mitigates the effects of the EEG electrode configuration, given the increased variance that can arise from using spatially harmonized EEG signals from both low- and high-density electrode configurations. Comparable dimensionality-reduction strategies have been applied in other dementia studies investigating brain oscillations in the EEG source space 52 , 96 . BAG estimation A linear regression model was constructed to estimate brain age from EEG features using Elastic Net 97 . This regularization technique controls model complexity by balancing feature selection and coefficient shrinkage. Penalties in the model promote sparsity by setting some coefficients to zero, enabling feature selection and shrinking correlated coefficients toward similar values, thereby mitigating multicollinearity and overfitting. Elastic Net is advantageous for assessing high-dimensional datasets with correlated predictors 98 , thereby ensuring robustness and interpretability in brain age prediction. Brain ages were computed using bagging, an ensemble learning technique that enhances predictive accuracy and reduces variance by training multiple models on bootstrapped subsets of the data and averaging their outputs 99 . Bagging is an ensemble learning method that generates multiple models by training on different subsets of the data and then averaging the predictions. In this study, 20 brain age estimates were obtained for each participant and subsequently averaged to enhance reliability. For each data subset, the HC sample was split into 90% and 10% for training and testing, respectively 100 . The model was trained using 10-fold cross-validation (Fig. 3E ). Furthermore, the regression model obtained during each cross-validation iteration was used to estimate the brain age of one-tenth of the individuals in each clinical group. This process was repeated 20 times, with the data randomly shuffled in each iteration 100 . Consequently, 20 brain age estimates were generated for each participant, with the mean serving as the final brain age prediction (Fig. 3E ). The BAG was calculated as the difference between each individual’s actual chronological age and the mean brain age prediction. A positive BAG (BAG > 0) indicated that the brain’s functional age appeared older than expected for the individual’s chronological age. Conversely, a negative BAG (BAG < 0) indicated that the brain’s functional attributes resembled those of younger peers (Fig. 3F ). A bias systematically exhibited by brain age models is the overestimation and underestimation of brain age in younger and older individuals, respectively, such that brain age is more accurately predicted for participants with a chronological age closer to the mean age of the training dataset 101 . This regression-toward-the-mean bias was adjusted in each cross-validation fold by linearly regressing the chronological age of the training set on the BAG estimates and using the regression residuals to correct the BAG estimates. Subsequently, the regression coefficients were used to adjust the BAG for individuals comprising the test set 5 , 29 . For each cross-validation fold, we computed the coefficient of determination ( R 2 ), the mean absolute error (MAE), the mean squared error (MSE), Fisher’s F , and the model’s significance. These metrics were used to evaluate the model’s performance. Furthermore, SHapley Additive exPlanations (SHAP) values 34 were computed to evaluate the relative importance of each feature in the regression models (Fig. 3F ). BAG classifications An XGBoost classifier was used to distinguish HC participants with low and high BAG, using country-level structural inequality (measured by the Gini coefficient), demographic factors (sex and educational attainment), and cognitive status (MMSE score) as predictive variables. The BAG was binarized using the median (−0.09) rather than zero to ensure a balanced class distribution. A second XGBoost classifier was used to predict low versus high BAG across the entire sample as a function of the clinical diagnosis (i.e., HC, MCI, bvFTD, AD), demographic factors (sex and educational attainment), and cognitive status. For model training and evaluation, the data was split into 80% for training and 20% for testing. A k -fold ( k = 10) cross-validation was used in the training phase. Performance classification metrics were reported, including accuracy, precision, and recall. Statistics and reproducibility The study sample size (1319 participants) exceeded the minimum threshold required to detect a small effect size (Cohen f 2 = 0.04) in multiple linear regression analyses ( F test, fixed model) and a medium effect size (Cohen’s f =0.15) in fixed-effects ANOVA ( F test, main effects, and interactions), ensuring a statistical power of 0.95 102 . A Welch´s ANOVA ( p < 0.05) was performed to compare the mean BAG across groups and to examine variations in the BAG of HC as a function of structural inequality (Fig. 3A , Table 1 ). This parametric test was chosen due to its suitability for cases where the study has a large sample size, but the assumption of homoscedasticity is not met 103 . Furthermore, a two-way ANOVA ( p < 0.05) was conducted to analyze variations in BAG as a function of the structural inequality of the country of residence and participants’ clinical diagnosis. The factor “structural inequality” included two levels: middle and high Gini, while the factor “group” comprised HC and AD. Unfortunately, restricting the ANOVA factors to two levels was necessary due to the absence of a representative sample of AD patients from countries classified in the lower GINI group and the limited geographical distribution of MCI and bvFTD patients (Table 1 ). For example, the MCI sample from medium-Gini-level countries comprised only Turkish participants, while the high-Gini-level bvFTD sample included only Colombian participants. The Fisher LSD test ( p < 0.05) was the post hoc test for multiple comparisons in all statistical analyses. To analyze the effect of possible confounders, we assessed whether the number of electrodes across sites influenced the results by testing associations between the number of EEG electrodes and country-level structural inequality. Additionally, consistent with our previous studies 24 , 61 , the ODQ index 36 of the EEG signals estimated in source space was used to control for potential confounds related to electrode number, sample size, and structural inequality. The ODQ quantifies the proportion of high-quality epochs in a recording by integrating signal-quality parameters such as maximum amplitude, standard deviation, correlation coefficient, and signal-to-noise ratio. Values range from 0, indicating that all epochs were classified as low quality, to 100, indicating that all epochs were classified as high quality 36 . As in previous studies 24 , 61 , logistic regression and group-level statistical analyses were performed with the Gini coefficient as the dependent variable to evaluate whether ODQ, the number of participants per center, or the number of channels predicted structural inequality values. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information (398.1KB, pdf) Reporting Summary (2.8MB, pdf) Transparent Peer Review file (601.9KB, pdf) Acknowledgements M.O. discloses support for the research of this work from Agencia Nacional de Investigación y Desarrollo (ANID) BASAL FB210008, ANID FONDECYT Iniciación 11241484 and ANID EXPLORACION 13240042. H.H. discloses support for the research of this work from the Davos Alzheimer’s Collaborative (8502-2025). E.H. discloses support for the research of this work from the Interfaculty Grant of Universidad Icesi CA03130123. J.M.C. discloses support for the research of this work, Ikerbasque: The Basque Foundation for Science, and from the Spanish Ministry of Science (PID2023-148008OB-I00), Spanish Ministry of Health (PI22/01118) and Basque Ministry of Health (2025111091, 2023111002, 2022111031). M.A.P. discloses support for the research of this work from the BrainLat Seed Grant BL-SRGP2020-02. A.I. discloses support for the research of this work from the Multi-partner consortium to expand dementia research in Latin America (ReDLat2), supported by the Fogarty International Center (FIC), National Institutes of Health and National Institute on Aging (R01 AG057234, R01 AG075775, R01 AG21051, R01 AG083799, CARDS-NIH), the Alzheimer’s Association (SG-20-725707), the Rainwater Charitable Foundation (The Bluefield Project to Cure FTD), ANID FONDECYT Regular (1250091, 1210176, 1220995), ANID/PIA/ANILLOS (ACT210096), JPI JPND-Care (DISCeRN 2025), ANID FONDEF (ID20I10152), ANID/FONDAP (15150012), the Wellcome Trust (335293/Z/25/Z), the Wellcome Leap CARE Program (CARE-2025-0883490149), and Horizon Europe (Marie Skłodowska-Curie Actions; CliCBrain 101236426). P.P. discloses support for the research of this work from ANID FONDECYT Regular 1260530. The contents of this publication are solely the responsibility of the authors and do not represent the official views of these institutions. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. Author contributions M.O., A.I, and P.P. conceived the study. A.S. and J.C. curated the data. J.C. and F.I.C. conducted the analyses. H.H and J.G.C. supervised the code implementation and data analysis. H.H., P.P., and A.I. supervised the study. M.O., A.I., and P.P. interpreted the results. M.O. wrote the first draft. P.P. designed the figures. A.I. M.O., F.I.C., C.B, A.I., and P.P. made specific contributions to sections in the manuscript. All authors reviewed, edited, provided critical comments, and approved the manuscript before submission: M.O., F.I.C., H.H., J.C., J.G.C., A.S., H.S., A.L., A.B., S.F., F.L., J.O., D.Ag., A.G., J.B., R.A.G., G.G.Y., B.G., İ.K., T.A., E.Y., L.H., R.A., P.A.V., R.G., J.E., S.L., R.W., A.F., A.M.G., D.H., M.S., E.H., D.Ab., N.R., R.A.C., W.E., J.M.C., M.A.P., C.B., A.I., and P.P. Peer review Peer review information Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Sahba Besharati and Jasmine Pan. A peer review file is available. Data availability All relevant data are included in the manuscript figures and the supplementary information. Source data underlying all graphs and charts are deposited and publicly available at Figshare, under the project name Numerical source data for graphs alpha clock EEG 104 . Row EEG files are available upon request to the Euro-LAD EEG Consortium 26 . A portion of the EEG dataset is publicly available in the BrainLat repository 27 . Code availability The EEG processing pipeline is publicly available 81 , 83 . All scripts for computational analyses, including the brain age models and classifications, are released on GitHub ( https://github.com/CompNeuroLabUSS/Alpha_Brain_Clocks ). Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Change history 6/22/2026 In this version of the article initially published, details of Ebru Yıldırım affiliations were incorrect and have now been amended to: Neuroscience and Neurotechnology Center of Excellence (NÖROM), Gazi University, Ankara, Türkiye in the HTML and PDF versions of the article. Contributor Information Agustin Ibanez, Email: agustin.ibanez@gbhi.org. Pavel Prado, Email: pavel.prado@uss.cl. Supplementary information The online version contains supplementary material available at 10.1038/s42003-026-10205-z. References 1. Gilbert, S. F. Developmental Biology (Sinauer Associates, 2000). 2. Khan, S. S., Singer, B. D. & Vaughan, D. E. Molecular and physiological manifestations and measurement of aging in humans. Aging Cell 16 , 624–633 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Cole, J. H. et al. Brain age predicts mortality. Mol. Psychiatry 23 , 1385–1392 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Gonneaud, J. et al. Accelerated functional brain aging in pre-clinical familial Alzheimer’s disease. Nat. Commun. 12 , 5346 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Moguilner, S. et al. Brain clocks capture diversity and disparities in aging and dementia across geographically diverse populations. Nat. Med. 10.1038/s41591-024-03209-x (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Bashyam, V. M. et al. MRI signatures of brain age and disease over the lifespan based on a deep brain network and 14 468 individuals worldwide. Brain 143 , 2312–2324 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Baecker, L. et al. Brain age prediction: a comparison between machine learning models using region- and voxel-based morphometric data. Hum. Brain Mapp. 42 , 2332–2346 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Chen, C.-L. et al. Generalization of diffusion magnetic resonance imaging-based brain age prediction model through transfer learning. NeuroImage 217 , 116831 (2020). [ DOI ] [ PubMed ] [ Google Scholar ] 9. Garcia Condado, J., Cortes, J. M. & for the Alzheimer’s Disease Neuroimaging Initiative NeuropsychBrainAge: a biomarker for conversion from mild cognitive impairment to Alzheimer’s disease. Alzheimer's Dement. 15 , e12493 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Coronel-Oliveros, C. et al. Creative experiences and brain clocks. Nat. Commun. 16 , 8336 (2025). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Coronel-Oliveros, C. et al. Diversity-sensitive brain clocks linked to biophysical mechanisms in aging and dementia. Nat. Mental Health 3 , 1214–1229 (2025). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Millar, P. R. et al. Predicting brain age from functional connectivity in symptomatic and preclinical Alzheimer disease. NeuroImage 256 , 119228 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Engemann, D. A. et al. A reusable benchmark of brain-age prediction from M/EEG resting-state signals. NeuroImage 262 , 119521 (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 14. Engemann, D. A. et al. Combining magnetoencephalography with magnetic resonance imaging enhances learning of surrogate-biomarkers. eLife 9 , e54055 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Babiloni, C. et al. Alpha rhythm and Alzheimer’s disease: has Hans Berger’s dream come true? Clinc. Neurophysiol. 172 , 33–50 (2025). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Aurlien, H. et al. EEG background activity described by a large computerized database. Clin. Neurophysiol. 115 , 665–673 (2004). [ DOI ] [ PubMed ] [ Google Scholar ] 17. Tröndle, M. et al. Decomposing age effects in EEG alpha power. Cortex 161 , 116–144 (2023). [ DOI ] [ PubMed ] [ Google Scholar ] 18. Chu, K.-T. et al. A holo-spectral EEG analysis provides an early detection of cognitive decline and predicts the progression to Alzheimer’s disease. Front. Aging Neurosci. 15 , 1195424 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Babiloni, C. et al. Sources of cortical rhythms in adults during physiological aging: a multicentric EEG study. Hum. Brain Mapp. 27 , 162–172 (2006). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Merkin, A. et al. Do age-related differences in aperiodic neural activity explain differences in resting EEG alpha? Neurobiol. Aging 121 , 78–87 (2023). [ DOI ] [ PubMed ] [ Google Scholar ] 21. Babiloni, C. et al. Resting state cortical rhythms in mild cognitive impairment and Alzheimer’s disease: electroencephalographic evidence. JAD . 26 , 201–214 (2011). [ DOI ] [ PubMed ] [ Google Scholar ] 22. Lejko, N., Larabi, D. I., Herrmann, C. S., Aleman, A. & Ćurčić-Blake, B. Alpha power and functional connectivity in cognitive decline: a systematic review and meta-analysis. J. Alzheimers Dis. 78 , 1047–1088 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Babiloni, C. et al. Abnormalities of resting state cortical EEG rhythms in subjects with mild cognitive impairment due to Alzheimer’s and Lewy body diseases. J. Alzheimers Dis. 62 , 247–268 (2018). [ DOI ] [ PubMed ] [ Google Scholar ] 24. Hernandez, H. et al. Brain health in diverse settings: how age, demographics and cognition shape brain function. Neuroimage 295 , 120636 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Kopčanová, M. et al. Resting-state EEG signatures of Alzheimer’s disease are driven by periodic but not aperiodic changes. Neurobiol. Dis. 190 , 106380 (2024). [ DOI ] [ PubMed ] [ Google Scholar ] 26. Parra-Rodriguez, M. A. et al. The EuroLaD-EEG consortium: towards a global EEG platform for dementia, for seeking to reduce the regional impact of dementia. Alzheimer’s Dement. 18 , e059944 (2022). 27. Prado, P. et al. The BrainLat project, a multimodal neuroimaging dataset of neurodegeneration from underrepresented backgrounds. Sci. Data 10 , 889 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Baez, S. et al. Structural inequality and temporal brain dynamics across diverse samples. Clin. Transl. Med. 14 , e70032 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Boyle, P. A. et al. The “cognitive clock”: a novel indicator of brain health. Alzheimer’s Dement. 17 , 1923–1937 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Klimesch, W. EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. Brain Res. Brain Res. Rev. 29 , 169–195 (1999). [ DOI ] [ PubMed ] [ Google Scholar ] 31. Bazanova, O. M. & Vernon, D. Interpreting EEG alpha activity. Neurosci. Biobehav. Rev. 44 , 94–110 (2014). [ DOI ] [ PubMed ] [ Google Scholar ] 32. Moretti, D. V. et al. Individual analysis of EEG frequency and band power in mild Alzheimer’s disease. Clin. Neurophysiol. 115 , 299–308 (2004). [ DOI ] [ PubMed ] [ Google Scholar ] 33. Babiloni, C. et al. International Federation of Clinical Neurophysiology (IFCN)—EEG research workgroup: recommendations on frequency and topographic analysis of resting state EEG rhythms. Part 1: applications in clinical research studies. Clin. Neurophysiol. 131 , 285–307 (2020). [ DOI ] [ PubMed ] [ Google Scholar ] 34. Lundberg, S. M. & Lee, S.-I. A unified approach to interpreting model predictions. In Advances in Neural Information Processing Systems Vol. 30 (eds Guyon, I. et al.) (Curran Associates, Inc., 2017). 35. Gini, C. Variabilità e Mutabilità: Contributo Allo Studio Delle Distribuzioni e Delle Relazioni Statistiche. [Fasc. I.] (Tipografia di Paolo Cuppini, 1912). 36. Zhao, L. et al. Quantitative signal quality assessment for large-scale continuous scalp electroencephalography from a big data perspective. Physiol. Meas. 44 , 035009 (2023). [ DOI ] [ PubMed ] [ Google Scholar ] 37. Folstein, M. F., Folstein, S. E. & McHugh, P. R. Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12 , 189–198 (1975). [ DOI ] [ PubMed ] [ Google Scholar ] 38. Al Zoubi, O. et al. Predicting age from brain EEG signals—a machine learning approach. Front. Aging Neurosci. 10 , 184 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Yook, S. et al. Novel neuroelectrophysiological age index associated with imaging features of brain aging and sleep disorders. Neuroimage 264 , 119753 (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 40. Kim, S. et al. Development of efficient brain age estimation method based on regional brain volume from structural magnetic resonance imaging. Psychiatry Investig. 21 , 37–43 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Xue, L. et al. 18F]FDG PET integrated with structural MRI for accurate brain age prediction. Eur. J. Nucl. Med. Mol. Imaging 51 , 3617–3629 (2024). [ DOI ] [ PubMed ] [ Google Scholar ] 42. Yook, S., Park, H. R., Joo, E. Y. & Kim, H. Predicting the impact of CPAP on brain health: a study using the sleep EEG -derived brain age index. Ann. Clin. Transl. Neurol. 11 , 1172–1183 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Babiloni, C. et al. Resting state cortical electroencephalographic rhythms are related to gray matter volume in subjects with mild cognitive impairment and Alzheimer’s disease. Hum. Brain Mapp. 34 , 1427–1446 (2013). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 44. Babiloni, C. et al. White-matter lesions along the cholinergic tracts are related to cortical sources of EEG rhythms in amnesic mild cognitive impairment. Hum. Brain Mapp. 30 , 1431–1443 (2009). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Lopez, S. et al. The association between posterior resting-state EEG alpha rhythms and functional MRI connectivity in older adults with subjective memory complaint. Neurobiol. Aging 137 , 62–77 (2024). [ DOI ] [ PubMed ] [ Google Scholar ] 46. Lőrincz, M. L., Kékesi, K. A., Juhász, G., Crunelli, V. & Hughes, S. W. Temporal framing of thalamic relay-mode firing by phasic inhibition during the alpha rhythm. Neuron 63 , 683–696 (2009). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Ricceri, L. et al. Cognitive and neurological deficits induced by early and prolonged basal forebrain cholinergic hypofunction in rats. Exp. Neurol. 189 , 162–172 (2004). [ DOI ] [ PubMed ] [ Google Scholar ] 48. Babiloni, C. et al. Abnormalities of cortical neural synchronization mechanisms in subjects with mild cognitive impairment due to Alzheimer’s and Parkinson’s diseases: an EEG study. J. Alzheimers Dis. 59 , 339–358 (2017). [ DOI ] [ PubMed ] [ Google Scholar ] 49. Klimesch, W., Doppelmayr, M., Russegger, H., Pachinger, T. & Schwaiger, J. Induced alpha band power changes in the human EEG and attention. Neurosci. Lett. 244 , 73–76 (1998). [ DOI ] [ PubMed ] [ Google Scholar ] 50. Babiloni, C. et al. Resting state electroencephalographic alpha rhythms are sensitive to Alzheimer’s disease mild cognitive impairment progression at a 6-month follow-up. Neurobiol. Aging 137 , 19–37 (2024). [ DOI ] [ PubMed ] [ Google Scholar ] 51. Babiloni, C. et al. Mapping distributed sources of cortical rhythms in mild Alzheimer’s disease. A multicentric EEG study. NeuroImage 22 , 57–67 (2004). [ DOI ] [ PubMed ] [ Google Scholar ] 52. Babiloni, C. et al. Brain neural synchronization and functional coupling in Alzheimer’s disease as revealed by resting state EEG rhythms. Int. J. Psychophysiol. 103 , 88–102 (2016). [ DOI ] [ PubMed ] [ Google Scholar ] 53. Babiloni, C. et al. Occipital sources of resting-state alpha rhythms are related to local gray matter density in subjects with amnesic mild cognitive impairment and Alzheimer’s disease. Neurobiol. Aging 36 , 556–570 (2015). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 54. Moretti, D. V. et al. Volumetric differences in mapped hippocampal regions correlate with increase of high alpha rhythm in Alzheimer’s disease. Int. J. Alzheimers Dis. 2011 , 208218 (2011). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 55. Moretti, D. V., Paternicò, D., Binetti, G., Zanetti, O. & Frisoni, G. B. EEG upper/low alpha frequency power ratio relates to temporo-parietal brain atrophy and memory performances in mild cognitive impairment. Front. Aging Neurosci. 5 , 63 (2013). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 56. Otero, M., Lea-Carnall, C., Prado, P., Escobar, M.-J. & El-Deredy, W. Modelling neural entrainment and its persistence: influence of frequency of stimulation and phase at the stimulus offset. Biomed. Phys. Eng. Express 8 , 045014 (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 57. Jones, B. F. et al. Differential regional atrophy of the cingulate gyrus in Alzheimer Disease: a volumetric MRI study. Cereb. Cortex 16 , 1701–1708 (2005). [ DOI ] [ PubMed ] [ Google Scholar ] 58. Babiloni, C. et al. Relationship between default mode network and resting-state electroencephalographic alpha rhythms in cognitively unimpaired seniors and patients with dementia due to Alzheimer’s disease. Cereb. Cortex 33 , 10514–10527 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 59. Hafkemeijer, A. et al. Resting state functional connectivity differences between behavioral variant frontotemporal dementia and Alzheimer’s disease. Front. Hum. Neurosci . 9 , 474 (2015). [ DOI ] [ PMC free article ] [ PubMed ] 60. Seeley, W. W. The Salience Network: a neural system for perceiving and responding to homeostatic demands. J. Neurosci. 39 , 9878–9882 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 61. Legaz, A. et al. Structural inequality linked to brain volume and network dynamics in aging and dementia across the Americas. Nat. Aging 5 , 259–274 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 62. Fittipaldi, S. et al. Heterogeneous factors influence social cognition across diverse settings in brain health and age-related diseases. Nat. Mental Health 2 , 63–75 (2024). [ Google Scholar ] 63. Amoruso, L. et al. Multilingualism protects against accelerated aging in cross-sectional and longitudinal analyses of 27 European countries. Nat. Aging 5 , 2340–2354 (2025). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 64. Vidal-Pineiro, D. et al. Individual variations in ‘brain age’ relate to early-life factors more than to longitudinal brain change. eLife 10 , e69995 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 65. Hammers, D. B. et al. Differences in baseline cognitive performance between participants with early-onset and late-onset Alzheimer’s disease: Comparison of LEADS and ADNI. Alzheimer’s Dement. 21 , e14218 (2025). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 66. Kothe, C. A. & Makeig, S. BCILAB: a platform for brain–computer interface development. J. Neural Eng. 10 , 056014 (2013). [ DOI ] [ PubMed ] [ Google Scholar ] 67. Delorme, A. & Makeig, S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J. Neurosci. Methods 134 , 9–21 (2004). [ DOI ] [ PubMed ] [ Google Scholar ] 68. Farzan, F. et al. Standardization of electroencephalography for multi-site, multi-platform and multi-investigator studies: insights from the Canadian Biomarker Integration Network in Depression. Sci. Rep. 7 , 7473 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 69. Bigdely-Shamlo, N. et al. Automated EEG mega-analysis II: cognitive aspects of event related features. NeuroImage 207 , 116054 (2020). [ DOI ] [ PubMed ] [ Google Scholar ] 70. Birba, A. et al. Allostatic-interoceptive overload in frontotemporal dementia. Biol. Psychiatry 92 , 54–67 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 71. Herzog, R. et al. Genuine high-order interactions in brain networks and neurodegeneration. Neurobiol. Disease 175 , 105918 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 72. Cruzat, J. et al. Temporal irreversibility of large-scale brain dynamics in Alzheimer’s disease. J. Neurosci. 43 , 1643–1656 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 73. Prado, P. et al. Source space connectomics of neurodegeneration: one-metric approach does not fit all. Neurobiol. Disease 179 , 106047 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 74. Legaz, A. et al. Social and non-social working memory in neurodegeneration. Neurobiol. Disease 183 , 106171 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 75. Coronel-Oliveros, C. et al. Viscous dynamics associated with hypoexcitation and structural disintegration in neurodegeneration via generative whole-brain modeling. Alzheimers Dement. 20 , 3228–3250 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 76. WorldBank. Poverty and Inequality Platform Methodology Handbook https://datanalytics.worldbank.org/PIP-Methodology/ (2025). 77. Creavin, S. T. et al. Mini-Mental State Examination (MMSE) for the detection of dementia in clinically unevaluated people aged 65 and over in community and primary care populations. Cochrane Database Syst. Rev. 2016 , CD011145 (2016). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 78. Petersen, R. C. Mild cognitive impairment as a diagnostic entity. J. Intern. Med. 256 , 183–194 (2004). [ DOI ] [ PubMed ] [ Google Scholar ] 79. McKhann, G. M. et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s Dement. 7 , 263–269 (2011). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 80. Rascovsky, K. et al. Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia. Brain 134 , 2456–2477 (2011). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 81. Prado, P. et al. Harmonized multi-metric and multi-centric assessment of EEG source space connectivity for dementia characterization. Alzheimer's Dement. 15 , e12455 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 82. Prado, P. et al. Dementia ConnEEGtome: towards multicentric harmonization of EEG connectivity in neurodegeneration. Int. J. Psychophysiol. 172 , 24–38 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 83. Sainz-Ballesteros, A., Perez, J. A. M., Moguilner, S., Ibáñez, A. & Prado, P. A pipeline for large-scale assessments of dementia EEG connectivity across multicentric settings. In Methods for Analyzing Large Neuroimaging Datasets Vol. 218 (eds Whelan, R. & Lemaître, H.) 229–253 (Springer US, 2025). 84. Dong, L. et al. MATLAB toolboxes for reference electrode standardization technique (REST) of scalp EEG. Front. Neurosci. 11 , 601 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 85. Pion-Tonachini, L., Kreutz-Delgado, K. & Makeig, S. ICLabel: an automated electroencephalographic independent component classifier, dataset, and website. NeuroImage 198 , 181–197 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 86. Bigdely-Shamlo, N., Kreutz-Delgado, K., Kothe, C. & Makeig, S. EyeCatch: data-mining over half a million EEG independent components to construct a fully-automated eye-component detector. 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) 5845–5848 (2013). [ DOI ] [ PMC free article ] [ PubMed ] 87. Melnik, A. et al. Systems, subjects, sessions: to what extent do these factors influence EEG data? Front. Hum. Neurosci. 11 , 150 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 88. Pascual-Marqui, R. D., Michel, C. M. & Lehmann, D. Low resolution electromagnetic tomography: a new method for localizing electrical activity in the brain. Int. J. Psychophysiol. 18 , 49–65 (1994). [ DOI ] [ PubMed ] [ Google Scholar ] 89. Pascual-Marqui, R. D. Standardized low-resolution brain electromagnetic tomography (sLORETA): technical details. Methods Find. Exp. Clin. Pharmacol. 24 (Suppl. D), 5–12 (2002). [ PubMed ] [ Google Scholar ] 90. Rolls, E. T., Joliot, M. & Tzourio-Mazoyer, N. Implementation of a new parcellation of the orbitofrontal cortex in the automated anatomical labeling atlas. NeuroImage 122 , 1–5 (2015). [ DOI ] [ PubMed ] [ Google Scholar ] 91. Martinez-Montes, E., Garcia-Puente, Y., Zanartu, M. & Prado-Gutierrez, P. Chirp analyzer for estimating amplitude and latency of steady-state auditory envelope following responses. IEEE Trans. Neural Syst. Rehabil. Eng. 28 , 2744–2753 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 92. Prado-Gutierrez, P. et al. Maturational time course of the envelope following response to amplitude-modulated acoustic signals in rats. Int. J. Audiol. 51 , 309–316 (2012). [ DOI ] [ PubMed ] [ Google Scholar ] 93. Wang, R. et al. Power spectral density and coherence analysis of Alzheimer’s EEG. Cogn. Neurodyn. 9 , 291–304 (2015). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 94. Jeong, H. T., Youn, Y. C., Sung, H.-H. & Kim, S. Y. Power spectral changes of quantitative EEG in the subjective cognitive decline: comparison of Community Normal Control Groups. Neuropsychiatr. Dis. Treatment 17 , 2783–2790 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 95. Capilla, A. et al. The natural frequencies of the resting human brain: an MEG-based atlas. NeuroImage 258 , 119373 (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 96. Lopez, S. et al. Patients with Alzheimer’s disease dementia show partially preserved parietal ‘hubs’ modeled from resting-state alpha electroencephalographic rhythms. Front. Aging Neurosci. 15 , 780014 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 97. Zou, H. & Hastie, T. Regularization and variable selection via the Elastic Net. J. R. Stat. Soc. Ser. B 67 , 301–320 (2005). [ Google Scholar ] 98. James, G., Witten, D., Hastie, T. & Tibshirani, R. An Introduction to Statistical Learning: With Applications in R (Springer US, 2021). 99. Tian, Y. E. et al. Heterogeneous aging across multiple organ systems and prediction of chronic disease and mortality. Nat. Med. 29 , 1221–1231 (2023). [ DOI ] [ PubMed ] [ Google Scholar ] 100. Dobbin, K. K. & Simon, R. M. Optimally splitting cases for training and testing high dimensional classifiers. BMC Med. Genom. 4 , 31 (2011). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 101. de Lange, A.-M. G. & Cole, J. H. Commentary: Correction procedures in brain-age prediction. Neuroimage Clin. 26 , 102229 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 102. Faul, F., Erdfelder, E., Lang, A.-G. & Buchner, A. G. Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav. Res. Methods 39 , 175–191 (2007). [ DOI ] [ PubMed ] [ Google Scholar ] 103. Moder, K. Alternatives to F -test in one way ANOVA in case of heterogeneity of variances (a simulation study). Psychol. Test Assess. Model. 52 , 343 (2010). [ Google Scholar ] 104. Prado, P. Numerical source data for graphs alpha clock EEG. 41661 Bytes figshare 10.6084/M9.FIGSHARE.31918965 (2026). Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary Information (398.1KB, pdf) Reporting Summary (2.8MB, pdf) Transparent Peer Review file (601.9KB, pdf) Data Availability Statement All relevant data are included in the manuscript figures and the supplementary information. Source data underlying all graphs and charts are deposited and publicly available at Figshare, under the project name Numerical source data for graphs alpha clock EEG 104 . Row EEG files are available upon request to the Euro-LAD EEG Consortium 26 . A portion of the EEG dataset is publicly available in the BrainLat repository 27 . The EEG processing pipeline is publicly available 81 , 83 . All scripts for computational analyses, including the brain age models and classifications, are released on GitHub ( https://github.com/CompNeuroLabUSS/Alpha_Brain_Clocks ). Articles from Communications Biology are provided here courtesy of Nature Publishing Group ACTIONS View on publisher site PDF (2.3 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Source-space EEG alpha activity reveals brain age gaps due to neurodegeneration and disparity#336285
• Data-driven lifespan transitions: cortical morphometry and intrinsic differences across network scales#230250
• Quantifying intergenerational differences in community public space perception: a multimodal Electroencephalography-eye tracking study.#30697
• Relating ADHD to the aperiodic component of the EEG signal in both males and females across development.#213909
• Early Adulthood Signatures of Motherhood in Brain Aging.#342365