Study: Brain clocks capture the diversity and differences in aging and dementia within geographically diverse populations. Image credits: Lightspring / Shutterstock
This is evident from a recent study published in the journal Naturopathy, researchers used deep learning to analyze the impact of geographical, socio-demographic, socio-economic, neurodegeneration-related and gender diversity on brain age differences in 15 countries. They found that structural socio-economic inequality, pollution and health disparities are important predictors of greater differences in brain age, especially in Latin American and Caribbean regions (LAC), with greater differences observed in women and individuals with cognitive disorders such as Alzheimer’s disease (AD). .
Background
The brain undergoes dynamic changes with age, which are crucial for understanding, especially regarding inequalities and brain disorders such as AD. Brain age models, which measure brain health based on several factors, have the potential to capture diversity in aging, but are underrepresented in underrepresented populations such as those in LAC. These populations face significant socioeconomic and health disparities, which may impact brain aging. Brain aging research has mainly focused on populations from the Global North and often uses structural magnetic resonance imaging (MRI), neglecting brain network dynamics captured by functional MRI (fMRI) and electroencephalograms (EEG). Although EEG is a more accessible tool in resource-limited settings, its use in large-scale studies is limited by challenges in standardization and integration with fMRI. There is a need to develop scalable brain age markers using deep learning that integrate these techniques and take into account demographic diversity, especially in underrepresented populations. Therefore, in the current study, researchers used graph convolutional networks to predict gaps in brain age and investigate the impact of diversity, including geographic, sociodemographic, and health factors, on brain aging.
About the study
The study analyzed resting-state fMRI and EEG datasets from 5,306 participants in 15 countries in the LAC and non-LAC regions. fMRI data was collected from 2,953 participants in Argentina, Chile, Colombia, Mexico, Peru, United States of America, China and Japan, while EEG data was collected from 2,353 participants in Argentina, Greece, Brazil, Chile, Colombia, Cuba , Ireland, Italy, Turkey and the United Kingdom. Participants included 3,509 healthy controls and 1,808 with neurocognitive disorders, namely mild cognitive impairment (MCI), AD or behavioral variant frontotemporal dementia (bvFTD). The data underwent rigorous preprocessing including normalization, noise correction, and source space estimation. High-order interactions between brain regions were assessed, with data converted into graphs for analysis via graph convolutional networks (GCNs). An approach with 80% cross-validation and 20% hold-out testing was used. Data augmentation techniques were used and the predictive performance of the model was evaluated using the goodness of fit (R²) and the root mean square error (rmse). Gradient boosting models were used to investigate the influence of exposome factors on brain age gaps. Extensive statistical analyzes were performed to validate the findings, including permutation tests and bootstrapping. The quality of the data was carefully assessed and the study followed strict ethical guidelines.
Datasets include LAC and non-LAC healthy controls (HC, total n = 3,509) and participants with Alzheimer’s disease (AD, total n = 828), bvFTD (total n = 463) and MCI (total n = 517) . The fMRI dataset included 2,953 participants from LAC (Argentina, Chile, Colombia, Mexico and Peru) and non-LAC (the US, China and Japan). The EEG dataset involved 2,353 participants from Argentina, Brazil, Chile, Colombia and Cuba (LAC), as well as Greece, Ireland, Italy, Turkey and the UK (non-LAC). The raw fMRI and EEG signals were preprocessed by filtering and artifact removal, and the EEG signals were normalized to project them into the source space. A parcellation using the automated anatomical labeling (AAL) atlas for both the fMRI and EEG signals was performed to build the nodes from which we calculated the high-order interactions using the Ω information metric. A connectivity matrix was obtained for both modalities, which was later graphed. Data augmentation was only performed in the test dataset. The graphs were used as input to a convolutional deep learning network (architecture shown in the last row), with separate models for EEG and fMRI. Finally, an age prediction was obtained and performance was measured by comparing the predicted ages with the chronological ages. This figure was created in part with BioRender.com (fMRI and EEG devices).
Results and discussion
The brain aging models showed adequate predictive performance. The most important predictive brain regional features were centered around frontoposterior networks, including nodes in the precentral gyrus, middle occipital gyrus, and superior and middle frontal gyri. Additional key nodes for the fMRI model were the inferior frontal gyri, the anterior and median cingulate, and the paracingulate gyri. For the EEG model, the inferior occipital gyrus and the superior and inferior parietal gyri were also significant.
Notably, when analyzing non-LAC datasets, the models showed similar patterns in predictive features, but with slightly reduced fit. In contrast, models trained on LAC datasets showed moderate fit and increased rmse values, highlighting biases in predicting older brain ages, especially for female participants. Furthermore, the study of cross-region effects showed that training on non-LAC data and testing on LAC resulted in positive mean directional errors (MDE), indicating biases towards older brain ages. Furthermore, brain age differences were observed to widen in clinical populations, suggesting accelerated aging in conditions such as MCI and AD compared to healthy controls.
These findings highlight the complexity of brain aging in different populations. They emphasize the importance of considering diversity factors in neurocognitive assessments. The study is enhanced by using diverse datasets from different countries, integrating fMRI and EEG data, and developing scalable, personalized brain health metrics applicable to diverse and underrepresented populations. However, research is limited by the lack of clinical EEG data from non-LAC regions, the reliance on unimodal measures of brain age, limited regional data, and the absence of individual-level demographic factors such as gender identity, socioeconomic status, and ethnicity. .
Conclusion
In conclusion, the study shows that brain clock models are sensitive to various factors such as geography, gender, macrosocial influences and diseases, despite the variability of the data. Using deep learning about high-order brain interactions via fMRI and EEG, the study paves the way for inclusive, accessible tools to assess differences in brain aging. It could potentially aid in the identification and staging of neurocognitive disorders such as MCI, AD and bvFTD and support personalized medicine approaches worldwide.