Brain–phenotype models fail for individuals who defy sample stereotypes

Nature, Journal Year: 2022, Volume and Issue: 609(7925), P. 109 - 118

Published: Aug. 24, 2022

Abstract Individual differences in brain functional organization track a range of traits, symptoms and behaviours 1–12 . So far, work modelling linear brain–phenotype relationships has assumed that single such relationship generalizes across all individuals, but models do not equally well participants 13,14 A better understanding whom fail why is crucial to revealing robust, useful unbiased relationships. To this end, here we related activity phenotype using predictive models—trained tested on independent data ensure generalizability 15 —and examined model failure. We applied data-driven approach neurocognitive measures new, clinically demographically heterogeneous dataset, with the results replicated two independent, publicly available datasets 16,17 Across three datasets, find reflect unitary cognitive constructs, rather scores intertwined sociodemographic clinical covariates; is, stereotypical profiles, when individuals who defy them. Model failure reliable, specific generalizable datasets. Together, these highlight pitfalls one-size-fits-all effect biased phenotypic 18–20 interpretation utility resulting models. present framework address issues so may reveal neural circuits underlie phenotypes ultimately identify individualized targets for intervention.

Language: Английский

---

Confronting racially exclusionary practices in the acquisition and analyses of neuroimaging data

Nature Neuroscience, Journal Year: 2022, Volume and Issue: 26(1), P. 4 - 11

Published: Dec. 23, 2022

Language: Английский

---

Multivariate BWAS can be replicable with moderate sample sizes

Nature, Journal Year: 2023, Volume and Issue: 615(7951), P. E4 - E7

Published: March 8, 2023

Language: Английский

---

Evidence for embracing normative modeling

eLife, Journal Year: 2023, Volume and Issue: 12

Published: March 13, 2023

In this work, we expand the normative model repository introduced in Rutherford et al., 2022a to include models charting lifespan trajectories of structural surface area and brain functional connectivity, measured using two unique resting-state network atlases (Yeo-17 Smith-10), an updated online platform for transferring these new data sources. We showcase value with a head-to-head comparison between features output by modeling raw several benchmarking tasks: mass univariate group difference testing (schizophrenia versus control), classification regression (predicting general cognitive ability). Across all benchmarks, show advantage features, strongest statistically significant results demonstrated tasks. intend accessible resources facilitate wider adoption across neuroimaging community.

Language: Английский

---

Functional connectomics in depression: insights into therapies

Trends in Cognitive Sciences, Journal Year: 2023, Volume and Issue: 27(9), P. 814 - 832

Published: June 5, 2023

Language: Английский

---

Homotopic local-global parcellation of the human cerebral cortex from resting-state functional connectivity

NeuroImage, Journal Year: 2023, Volume and Issue: 273, P. 120010 - 120010

Published: March 12, 2023

Resting-state fMRI is commonly used to derive brain parcellations, which are widely for dimensionality reduction and interpreting human neuroscience studies. We previously developed a model that integrates local global approaches estimating areal-level cortical parcellations. The resulting local-global parcellations often referred as the Schaefer However, lack of homotopic correspondence between left right parcels has limited their use lateralization Here, we extend our previous Using resting-fMRI task-fMRI across diverse scanners, acquisition protocols, preprocessing demographics, show homogeneous while being more than five publicly available Furthermore, weaker correlations associated with greater in resting network organization, well language motor task activation. Finally, agree boundaries number areas estimated from histology visuotopic fMRI, capturing sub-areal (e.g., somatotopic visuotopic) features. Overall, these results suggest represent neurobiologically meaningful subdivisions cerebral cortex will be useful resource future Multi-resolution 1479 participants (https://github.com/ThomasYeoLab/CBIG/tree/master/stable_projects/brain_parcellation/Yan2023_homotopic).

Language: Английский

---

Bias in machine learning models can be significantly mitigated by careful training: Evidence from neuroimaging studies

Proceedings of the National Academy of Sciences, Journal Year: 2023, Volume and Issue: 120(6)

Published: Jan. 30, 2023

Despite the great promise that machine learning has offered in many fields of medicine, it also raised concerns about potential biases and poor generalization across genders, age distributions, races ethnicities, hospitals, data acquisition equipment protocols. In current study, context three brain diseases, we provide evidence which suggests when properly trained, models can generalize well diverse conditions do not necessarily suffer from bias. Specifically, by using multi-study magnetic resonance imaging consortia for diagnosing Alzheimer's disease, schizophrenia, autism spectrum disorder, find well-trained have a high area-under-the-curve (AUC) on subjects different subgroups pertaining to attributes such as gender, age, racial groups, clinical studies are unbiased under multiple fairness metrics demographic parity difference, equalized odds equal opportunity difference etc. We incorporate multi-source demographic, clinical, genetic factors cognitive scores unbiased. These better predictive AUC than those trained only with features but there situations these additional help.

Language: Английский

---

Data leakage inflates prediction performance in connectome-based machine learning models

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: Feb. 28, 2024

Abstract Predictive modeling is a central technique in neuroimaging to identify brain-behavior relationships and test their generalizability unseen data. However, data leakage undermines the validity of predictive models by breaching separation between training Leakage always an incorrect practice but still pervasive machine learning. Understanding its effects on can inform how affects existing literature. Here, we investigate five forms leakage–involving feature selection, covariate correction, dependence subjects–on functional structural connectome-based learning across four datasets three phenotypes. via selection repeated subjects drastically inflates prediction performance, whereas other have minor effects. Furthermore, small exacerbate leakage. Overall, our results illustrate variable underscore importance avoiding improve reproducibility modeling.

Language: Английский

---

Heterogeneous factors influence social cognition across diverse settings in brain health and age-related diseases

Nature Mental Health, Journal Year: 2024, Volume and Issue: 2(1), P. 63 - 75

Published: Jan. 2, 2024

Abstract Aging diminishes social cognition, and changes in this capacity can indicate brain diseases. However, the relative contribution of age, diagnosis reserve to especially among older adults global settings, remains unclear when considering other factors. Here, using a computational approach, we combined predictors cognition from diverse sample 1,063 across nine countries. Emotion recognition, mentalizing overall were predicted via support vector regressions various factors, including (subjective cognitive complaints, mild impairment, Alzheimer’s disease behavioral variant frontotemporal dementia), demographics, cognition/executive function, motion artifacts functional magnetic resonance imaging recordings. Higher cognitive/executive functions education ranked top predictors, outweighing reserve. Network connectivity did not show predictive values. The results challenge traditional interpretations age-related decline, patient–control differences associations emphasizing importance heterogeneous

Language: Английский

---

MRI economics: Balancing sample size and scan duration in brain wide association studies

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Feb. 18, 2024

Abstract A pervasive dilemma in neuroimaging is whether to prioritize sample size or scan time given fixed resources. Here, we systematically investigate this trade-off the context of brain-wide association studies (BWAS) using functional magnetic resonance imaging (fMRI). We find that total duration (sample × per participant) robustly explains individual-level phenotypic prediction accuracy via a logarithmic model, suggesting and are broadly interchangeable up 20-30 min data. However, returns diminish relative size, which explain with principled theoretical derivations. When accounting for overhead costs associated each participant (e.g., recruitment, non-imaging measures), many small-scale some large-scale BWAS might benefit from longer than typically assumed. These results generalize across domains, scanners, acquisition protocols, racial groups, mental disorders, age as well resting-state task-state connectivity. Overall, our study emphasizes importance time, ignored standard power calculations. Standard calculations maximize at expense can result sub-optimal accuracies inefficient use Our empirically informed reference available future design: WEB_APPLICATION_LINK

Language: Английский

---