Network-level enrichment provides a framework for biological interpretation of machine learning results

Network Neuroscience, Journal Year: 2024, Volume and Issue: 8(3), P. 762 - 790

Published: Jan. 1, 2024

Abstract Machine learning algorithms are increasingly being utilized to identify brain connectivity biomarkers linked behavioral and clinical outcomes. However, research often prioritizes prediction accuracy at the expense of biological interpretability, inconsistent implementation ML methods may hinder model accuracy. To address this, our paper introduces a network-level enrichment approach, which integrates system organization in context connectome-wide statistical analysis reveal links between behavior. demonstrate efficacy this we used linear support vector regression (LSVR) models examine relationship resting-state functional networks chronological age. We compared associations based on raw LSVR weights those produced from forward inverse models. Results indicated that not accounting for shared family variance inflated performance, k-best feature selection via Pearson correlation reduced reliability, deviated significant systems identified by Our findings offer crucial insights applying machine neuroimaging data, emphasizing value network interpretation.

Language: Английский

---

Task functional networks predict individual differences in the speed of emotional facial discrimination

NeuroImage, Journal Year: 2024, Volume and Issue: 297, P. 120715 - 120715

Published: June 28, 2024

Every individual experiences negative emotions, such as fear and anger, significantly influencing how external information is perceived processed. With the gradual rise in brain-behavior relationship studies, analyses investigating differences emotion processing a more objective measure response time (RT) remain unexplored. This study aims to address this gap by establishing that speed of facial discrimination can be predicted from whole-brain functional connectivity when participants were performing face task. Employing connectome predictive modeling (CPM) framework, we demonstrated young healthy adult group Human Connectome Project-Young Adults (HCP-YA) dataset Boston Adolescent Neuroimaging Depression Anxiety (BANDA) dataset. We identified distinct network contributions adolescent models. The highest represented brain networks involved model predictions included representations motor, visual association, salience, medial frontal networks. Conversely, models showed substantial cerebellum-frontoparietal interactions. Finally, observed despite successful within-dataset prediction adults adolescents, failed cross-dataset generalization. In conclusion, our shows emotional samples using their during processing. Future research needed derivation generalizable

Language: Английский

---

Task‐specific topology of brain networks supporting working memory and inhibition

Human Brain Mapping, Journal Year: 2024, Volume and Issue: 45(13)

Published: Sept. 1, 2024

Abstract Network neuroscience explores the brain's connectome, demonstrating that dynamic neural networks support cognitive functions. This study investigates how distinct abilities—working memory and inhibitory control—are supported by unique brain network configurations constructed estimating whole‐brain using mutual information. The involved 195 participants who completed Sternberg Item Recognition task Flanker tasks while undergoing electroencephalography recording. A mixed‐effects linear model analyzed influence of metrics on performance, considering individual differences task‐specific dynamics. findings indicate working control are associated with different attributes, relying distributed more segregated ones. Our analysis suggests both strong weak connections contribute to processes, potentially leading a stable control. indirectly theory intelligence, suggesting functional topology inherent various Nevertheless, we propose understanding variations in abilities requires recognizing shared processes within

Language: Английский

---

Classification and prediction of cognitive performance differences in older age based on brain network patterns using a machine learning approach

Network Neuroscience, Journal Year: 2022, Volume and Issue: 7(1), P. 122 - 147

Published: Aug. 31, 2022

Age-related cognitive decline varies greatly in healthy older adults, which may partly be explained by differences the functional architecture of brain networks. Resting-state connectivity (RSFC) derived network parameters as widely used markers describing this have even been successfully to support diagnosis neurodegenerative diseases. The current study aimed at examining whether these also useful classifying and predicting performance normally aging using machine learning (ML). Classifiability predictability global domain-specific from nodal network-level RSFC strength measures were examined adults 1000BRAINS (age range: 55-85 years). ML was systematically evaluated across different analytic choices a robust cross-validation scheme. Across analyses, classification did not exceed 60% accuracy for cognition. Prediction equally low with high mean absolute errors (MAEs ≥ 0.75) none variance (R2 ≤ 0.07) targets, feature sets, pipeline configurations. Current results highlight limited potential serve sole biomarker emphasize that cognition patterns challenging.

Language: Английский

---

Protocol for a randomized controlled trial of mindfulness-based stress reduction to improve attentional control in older adults (HealthyAgers trial)

BMC Geriatrics, Journal Year: 2022, Volume and Issue: 22(1)

Published: Aug. 13, 2022

Mindfulness meditation is a form of mind-body intervention that has increasing scientific support for its ability to reduce age-related declines in cognitive functioning, improve affective health, and strengthen the neural circuitry supporting improved health. However, majority existent studies have been pilot investigations with small sample sizes, limited follow-up data, lack attention expectancy effects. Here, we present study design Phase I/II, efficacy trial-HealthyAgers trial-that examines benefits manualized mindfulness-based stress reduction program improving attentional control reducing mind-wandering older adults.One hundred fifty adults (ages 65-85 years) will be randomized into one two groups: an eight-week mindfulness or eight-week, placebo-controlled, lifestyle education program. Behavioral neuroimaging assessments are conducted before after training. Participants then invited booster sessions once every three months period 12 post-intervention at 6-months 12-months. The primary outcomes behavioral measures mind-wandering. Additional, secondary include network strength priori defined neuromarker control, fluid everyday cognition, emotion regulation strategy use, markers inflammation.This establish group-based, low-cost inter-related facets adults. Strengths this well-designed, placebo-controlled comparison group, use web/mobile application track adherence, longitudinal follow-up.Clinicaltrials.gov (# NCT03626532 ). Registered August 4, 2018.

Language: Английский

---

Connectome-based predictive modeling of cognitive reserve using task-based functional connectivity

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2022, Volume and Issue: unknown

Published: June 2, 2022

Abstract Cognitive reserve supports cognitive function in the presence of pathology or atrophy. Functional neuroimaging may enable direct and accurate measurement which could have considerable clinical potential. The present study aimed to develop validate a measure using task-based fMRI data that then be applied independent resting-state data. Connectome-based predictive modeling with leave-one-out cross-validation was predict residual functional connectivity from Reserve/Reference Ability Neural Network studies (n = 220, mean age 51.91 years, SD 17.04 years). Three network-strength predicted measures were generated accurately unseen participants. theoretical validity these established via positive correlation socio-behavioural proxy (verbal intelligence) global cognition, brain structure. This fitted model external test data: Irish Longitudinal Study on Ageing (TILDA, n 294, 68.3 7.18 not positively associated nor verbal intelligence cognition. demonstrated can used generate theoretically valid reserve. Further work is needed establish if, how, derived

Language: Английский

---

Beyond network connectivity: A classification approach to brain age prediction with resting-state fMRI

NeuroImage, Journal Year: 2024, Volume and Issue: 290, P. 120570 - 120570

Published: March 11, 2024

The brain is a complex, dynamic organ that shows differences in the same subject at various periods. Understanding how activity changes across age as function of networks has been greatly abetted by fMRI. Canonical analysis consists determining alterations connectivity patterns (CPs) certain regions are affected. An alternative approach taken here not considering but rather features computed from recordings interest (ROIs). Using machine learning (ML) we assess neural signals altered and prospectively predictive sex via methodology novel drawing upon pairwise classification six decades subjects' chronological ages. ML used to answer equally important questions what properties most well which affected aging. It was found there decreased differentiation among older subjects separated number years younger subjects. Furthermore, burstiness change different rates between males females. findings provide insight into aging an ROI-based analysis, consideration several feature groups, classification-based pipeline. There also contribution understanding effects data aggregated recording centers on conclusions fMRI studies.

Language: Английский

---

Network level enrichment provides a framework for biological interpretation of machine learning results

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Oct. 17, 2023

Abstract Machine learning algorithms are increasingly used to identify brain connectivity biomarkers linked behavior and clinical outcomes. However, non-standard methodological choices in neuroimaging datasets, especially those with families or twins, have prevented robust machine applications. Additionally, prioritizing prediction accuracy over biological interpretability has made it challenging understand the processes behind psychopathology. In this study, we employed a linear support vector regression model study relationship between resting-state functional networks chronological age using data from Human Connectome Project. We examined effect of shared variance twins siblings by cross-validation, either randomly assigning keeping family members together. also compared models without Pearson feature filter utilized network enrichment approach predictive networks. Results indicated that not accounting for inflated performance, reduced reliability. Enhancing was achieved inverting applying network-level on connectome, while directly coefficients as weights led misleading interpretations. Our findings offer crucial insights data, emphasizing value comprehensible interpretation.

Language: Английский

---

TRACking health behaviors in people with Multiple Sclerosis (TRAC-MS): Study protocol and description of the study sample

Contemporary Clinical Trials Communications, Journal Year: 2022, Volume and Issue: 30, P. 101006 - 101006

Published: Sept. 20, 2022

People with multiple sclerosis (PwMS) experience a range of physical, cognitive, and affective symptoms. Behavioral interventions targeting increased physical activity show promising support as low-cost methods to improve working memory, episodic processing speed in PwMS. In this randomized controlled trial, we will examine the efficacy pedometer-tracking intervention, designed increase low-to-moderate levels activity, for improving memory

Language: Английский

---

Scene-selective regions encode the vertical position of navigationally relevant information in young and older adulthood

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Oct. 19, 2023

Abstract Position within the environment influences navigational relevance of objects. However, possibility that vertical position represents a central object property has yet to be explored. Considering upper and lower visual fields afford distinct types cues scene-selective regions exhibit retinotopic biases, it is interest elucidate whether location information modulates neural activity in these high-level areas. The occipital place area (OPA), parahippocampal (PPA) medial (MPA) demonstrate biases for contralateral field, hemifield, respectively. Interesting insights could also gained from studying older adulthood as recent work points towards an age-related preference field. In present study, young participants learned goal virtual manipulated two variables: navigationally-relevant objects presence non-relevant Results revealed all three parsed useful independently their subtending biases. It therefore appears representations higher-level system combined about value wayfinding purposes. This was maintained healthy aging emphasizing enduring significance processing along dimension spatial navigation abilities across lifespan.

Language: Английский

---