Predicting sex, age, general cognition and mental health with machine learning on brain structural connectomes

Yeung, Hon Wah; Stolicyn, Aleks; Buchanan, Colin R; Tucker‐Drob, Elliot M.; Bastin, Mark E.; Luz, Saturnino; McIntosh, Andrew M.; Whalley, Heather C.; Cox, Simon R.; Smith, Keith

doi:10.1002/hbm.26182

Cited by 11 publications

(5 citation statements)

References 112 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, to further evaluate the relevance of demographic variables in the prediction setting we investigated the individual contributions of age, sex, and education to the prediction by including these as extra features to the ML model. We found that the addition of age, sex, and education to our brain models drastically increased predictability of cognitive targets, in line with prior studies [31,32,55,106,107]. For example, Dadi et al showed that fluid intelligence and neuroticism were more successfully predicted when sociodemographic information was included into the model in a large sample from the UK Biobank (N = 11,175) [31].…”

Section: Discussionsupporting

confidence: 88%

Prediction of cognitive performance differences in older age from multimodal neuroimaging data

Krämer¹,

Stumme²,

Campos³

et al. 2023

GeroScience

View full text Add to dashboard Cite

Differences in brain structure and functional and structural network architecture have been found to partly explain cognitive performance differences in older ages. Thus, they may serve as potential markers for these differences. Initial unimodal studies, however, have reported mixed prediction results of selective cognitive variables based on these brain features using machine learning (ML). Thus, the aim of the current study was to investigate the general validity of cognitive performance prediction from imaging data in healthy older adults. In particular, the focus was with examining whether (1) multimodal information, i.e., region-wise grey matter volume (GMV), resting-state functional connectivity (RSFC), and structural connectivity (SC) estimates, may improve predictability of cognitive targets, (2) predictability differences arise for global cognition and distinct cognitive profiles, and (3) results generalize across different ML approaches in 594 healthy older adults (age range: 55–85 years) from the 1000BRAINS study. Prediction potential was examined for each modality and all multimodal combinations, with and without confound (i.e., age, education, and sex) regression across different analytic options, i.e., variations in algorithms, feature sets, and multimodal approaches (i.e., concatenation vs. stacking). Results showed that prediction performance differed considerably between deconfounding strategies. In the absence of demographic confounder control, successful prediction of cognitive performance could be observed across analytic choices. Combination of different modalities tended to marginally improve predictability of cognitive performance compared to single modalities. Importantly, all previously described effects vanished in the strict confounder control condition. Despite a small trend for a multimodal benefit, developing a biomarker for cognitive aging remains challenging.

show abstract

Section: Discussionsupporting

confidence: 88%

Prediction of cognitive performance differences in older age from multimodal neuroimaging data

Krämer¹,

Stumme²,

Campos³

et al. 2023

GeroScience

View full text Add to dashboard Cite

show abstract

“…We also found septum and fimbria among relevant nodes. Differences in these structures (Yeung, Stolicyn et al 2023) have been reported (Xin, Zhang et al 2019). We also found differences in hypothalamus, a well-known sexually dimorphic region (Scott, Prigge et al 2015).…”

Section: Discussionmentioning

confidence: 96%

“…the basal ganglia, hypothalamus (48[45], hippocampus, and amygdala [46]. Memory related circuits (including the entorhinal cortex, hippocampus, septum and fimbria [47] [48], were supplemented with cerebellum and gigantocellular nuclei circuits, suggesting differences in locomotor recovery after injury[39] [49].…”

Section: Discussionmentioning

confidence: 99%

APOE, Immune Factors, Sex, and Diet Interact to Shape Brain Networks in Mouse Models of Aging

Winter,

Mahzarnia,

Anderson

et al. 2023

Preprint

View full text Add to dashboard Cite

The exact mechanisms through which the APOE gene influences Alzheimers disease risk (AD) are under intense investigation. Yet, much remains unknown about how APOE alleles interact with age, sex, and diet to confer vulnerability to select brain networks. To address this question, we used mouse models that carry the three major human APOE alleles, and have a mouse like, or humanized innate immune system. This was realized through the replacement of the mouse with the human inducible nitric oxide synthase (iNOS) gene, leading to a lower immune response. Our study combines advanced, accelerated diffusion imaging methods with novel statistical analyses to reveal global and local brain graph differences, associated with each of the following risk factors for abnormal aging and AD: age, APOE genotype, sex, diet, innate immune response. We used a sparse logistic regression model (GraphClass) with complimentary pairs stability selection to identify small and robust subnetworks predictive of individual risk factors. The comparison of APOE3 versus APOE4 carriage identified 773 edges, with no more than 40 false selections; age resulted in 300 high selection probability edges with <5 false selections, sex resulted in 540 high selection probability edges with <7 false selections; diet in 1626 high selection probability edges with <31 false selections; and humanizing NOS in 411 edges with <9 false selections. Our results revealed widespread network differences due to age and diet, including the temporal association cortex, and also regions involved in response to threatening stimuli, such as the amygdala and periaqueductal gray. Sex associated differences affected regions such as the thalamus, insula and hypothalamus, but also fimbria and septum, involved in memory processes. APOE genotype was associated with differences in the connectivity of memory related areas, and also in sensory and motor areas; while diet and innate immunity (HN) were associated with differences in insula and hypothalamus connectivity. We pooled these models to identify common networks across multiple traits, giving insight into shared vulnerability amongst the risk factors. We identified 63 edges out of the total 54,946 edges (0.11% of the connectome) as common to all risk factors tested. Our results revealed common subnetworks vulnerable to several risk factors for AD, in an approach that can provide new biomarkers, and targets for therapies.

show abstract

“…They attributed this improvement to the ability of BrainNetCNN to extract hierarchical features from the functional connectome, including both global and local network properties. Another study also used BrainNetCNN for brain age prediction and found that it was able to capture nonlinear relationships between the functional connectivity patterns and brain age [27]. They suggested that this was due to the ability of BrainNetCNN to learn more complex and diverse features from the functional connectome.…”

Section: Discussionmentioning

confidence: 99%

Structure-decoupled functional connectome-based brain age prediction provides higher association to cognition

Chen,

Wang,

et al. 2023

NeuroReport

View full text Add to dashboard Cite

Brain age prediction as well as the prediction difference has been well examined to be a potential biomarker for brain disease or abnormal aging process. However, less knowledge was reported for the cognitive association within normal population. In this study, we proposed a novel approach to brain age prediction by structure-decoupled functional connectome. The original functional connectome was decomposed and decoupled into a structure-decoupled functional connectome using structural connectome harmonics. Our method was applied to a large dataset of normal aging individuals and achieved a high correlation between predicted and chronological age (r = 0.77). Both the original FC and structure-decoupled FC could be well-trained in a brain age prediction model. Significant remarkable relationships between the brain age prediction difference (predicted age minus chronological age) and cognitive scores were discovered. However, the brain age-predicted difference driven by structure-decoupled FC showed a stronger correction to the two cognitive scores (MMSE: r = −0.27, P-value = 0.002; MoCA: r = −0.32, P-value = 0.0003). Our findings suggest that our structure-decoupled functional connectivity approach could provide a more individual-specific functional network, leading to improved brain age prediction performance and a better understanding of cognitive decline in aging.

show abstract

Predicting sex, age, general cognition and mental health with machine learning on brain structural connectomes

Cited by 11 publications

References 112 publications

Prediction of cognitive performance differences in older age from multimodal neuroimaging data

Prediction of cognitive performance differences in older age from multimodal neuroimaging data

APOE, Immune Factors, Sex, and Diet Interact to Shape Brain Networks in Mouse Models of Aging

Structure-decoupled functional connectome-based brain age prediction provides higher association to cognition

Contact Info

Product

Resources

About