Multimodal brain-age prediction and cardiovascular risk: The Whitehall II MRI sub-study

Abstract

Brain age is becoming a widely applied imaging-based biomarker of neural aging and potential proxy for brain integrity and health. We estimated multimodal and modality-specific brain age in the Whitehall II MRI cohort using machine learning and imaging-derived measures of gray matter morphology, diffusion-based white matter microstructure, and resting state functional connectivity. Ten-fold cross validation yielded multimodal and modality-specific brain age estimates for each participant, and additional predictions based on a separate training sample was included for comparison. The results showed equivalent age prediction accuracy between the multimodal model and the gray and white matter models (R 2 of 0.34, 0.31, and 0.31, respectively), while the functional connectivity model showed a lower prediction accuracy (R 2 of 0.01). Cardiovascular risk factors, including high blood pressure, alcohol intake, and stroke risk score, were each associated with more apparent brain aging, with consistent associations across modalities.

The maternal brain: Region‐specific patterns of brain aging are traceable decades after childbirth

“…In line with recent brain‐age studies (de Lange et al, 2019; de Lange, Anatürk, et al, 2020; Kaufmann et al, 2019), the XGBoost regressor model , which is based on a decision‐tree ensemble algorithm (https://xgboost.readthedocs.io/en/latest/python), was used to estimate global and regional brain age based on the MRI data. XGboost includes advanced regularization to reduce overfitting (Chen & Guestrin, 2016), and uses a gradient boosting framework where the final model is based on a collection of individual models (https://github.com/dmlc/xgboost).…”

“…Brain‐age prediction was used to derive estimates of global brain aging, which was analyzed in relation to number of previous (live) childbirths in 8,880 newly added UK Biobank participants. Brain‐age prediction is commonly used to estimate an individual's age based on their brain characteristics (Cole & Franke, 2017), and individual variation in “brain age” estimates has been associated with a range of clinical and biological factors (Cole, 2020; Cole et al, 2017, 2018; Cole & Franke, 2017; Cole, Marioni, Harris, & Deary, 2019; de Lange, Anatürk, et al, 2020; de Lange, Barth, et al, 2020; Franke & Gaser, 2019; Kaufmann et al, 2019; Smith, Vidaurre, Alfaro‐Almagro, Nichols, & Miller, 2019). As compared to MRI‐derived measures such as cortical volume or thickness, brain‐age prediction adds a dimension by capturing deviations from normative aging trajectories identified by machine learning.…”

A history of previous childbirths is linked to women’s white matter brain age in midlife and older age

“…This estimation is then compared to the individual’s chronological age to estimate each individual’s brain-age gap (BAG), which is used to identify degrees of deviation from normative ageing trajectories. Such deviations have been associated with a range of clinical risk factors [37, 55, 61] as well as neurological and neuropsychiatric diseases [62, 63, 64, 36]. They have also been assessed in previous studies of parity and brain ageing [9, 10, 11, 54].…”

Prominent health problems, socioeconomic deprivation, and higher brain age in lonely and isolated individuals: A population-based study

“…In line with recent brain-age studies [29,33,34,6,28], the XGBoost regressor model, which aging trajectories [18]. The delta estimates were corrected for chronological age using linear regression [36,37].…”

“…Several factors may contribute to higher mortality risk in lonely individuals, including depression, adverse socioeconomic conditions, cardiovascular risk, and unhealthy lifestyle behaviors [3]. These risk factors are also linked to brain health [4,5,6,7,8], but the interplay between loneliness, brain health, and behavioral, socioeconomic, biological, and psychological risk factors is largely unexplored [9].…”