Pith. sign in

REVIEW 3 major objections 6 minor 55 references

A Brain Age Residual Biomarker (BARB): Leveraging MRI-Based Models to Detect Latent Health Conditions in U.S. Veterans

T0 review · 3 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper argues that the gap between MRI-predicted and chronological brain age—the model residual—tracks how many chronic health conditions a veteran carries, making it a candidate biomarker for latent disease.

desk verdict Decent brain-age model on an unusual veteran MRI dataset, but the headline residual–disease claim is confounded by age composition; the paper reports its own null ANOVA, so the honest core is the model, not the biomarker. read the letter →

arxiv 2501.05970 v2 pith:46YMRLJI submitted 2025-01-10 cs.LG

Shahrzad Jamshidi , Arthur Bousquet , Sugata Banerji , Mark F. Conneely , Bita Aslrousta This is my paper · ORCID
classification cs.LG
keywords brainagepredictionresidualbiomarkerconvolutionalneuralnetworkMRIT2-weightedimagingICDcodesveteranhealthlatentdiseasedetection
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that the residual of a brain-age model—the gap between the age a neural network predicts from MRI scans and the patient's chronological age—can serve as a biomarker for latent health conditions. Using four convolutional neural networks trained on T2-weighted FSE and FLAIR slices of 1,220 U.S. veterans, combined by a degree-3 polynomial ensemble, the authors predict age with an $R^2$ of 0.816 and then test whether residuals track five ICD-coded conditions: hypertension, diabetes, mild traumatic brain injury, and substance or alcohol abuse/dependence. They report that residuals grouped by the number of ICD codes show statistically significant different trend lines ($p = 0.002$), and that in veterans over 49, negative residuals—brains predicted to be older than they are—correlate with carrying multiple ICD codes. The point of the work is that a standard, non-invasive MRI sequence could flag otherwise hidden or underreported disease burden in a veteran population.

What carries the argument

The central object is the brain age residual biomarker (BARB), the difference between a model's predicted brain age and a patient's chronological age; a negative residual means the brain appears older than its owner's years. The machinery is a four-CNN ensemble: each of four identical shallow convolutional networks learns one of four 2D T2-weighted image types (FSE and FLAIR, at the anterior commissure and at the frontal horns of the lateral ventricles), and a degree-3 polynomial regression without interaction terms combines their outputs into the final age prediction. The residual is then grouped by number of ICD codes and analyzed through trend-line comparison and ANOVA, with the over-49 subgroup carrying the significant signal.

What would settle it

Re-run the residual comparison on age-matched subgroups: for each chronological age, compare residuals of veterans with zero, one, and multiple ICD codes. If the residual differences disappear within matched age bands—or if an age-adjusted residual ANOVA is null—the central claim is falsified.

Watch

Extended reading notes

Core claim

The central discovery, on the paper's own terms, is that the error of a brain-age predictor is not pure noise: it carries information about disease burden. The ensemble model, built from four lightweight CNNs each trained on one of four T2-weighted image types, produces residuals whose group-specific regression lines differ significantly by the number of ICD-coded conditions ($F = 4.265$, $p = 0.002$ in the training set; $F = 16.4$, $p = 5.77\times10^{-13}$ in the test set). The authors find the clearest signal in subjects older than 49, where a one-way ANOVA on residuals grouped by ICD count is significant ($p = 0.0069$) and negative residuals (older-appearing brain) accompany multiple diagnoses. They are careful to note that individual conditions, apart from hypertension and substance abuse/dependence in ANOVA, show no consistent association, and they attribute the aggregate signal to the collective burden of chronic conditions rather than to any single code.

Load-bearing premise

The load-bearing premise is that the different regression lines and negative residuals across ICD-count groups reflect disease effects on brain aging, not the fact that veterans with more diagnoses tend to be older and that prediction error grows with age.

Editorial extensions

If this is right

  • If the association holds, a gap between predicted and chronological brain age becomes a candidate screening signal for otherwise hidden disease burden in veterans, especially after age 49.
  • The success of the model on two T2-weighted sequences rather than T1 means routine clinical MRIs already collected for other reasons could be reused for brain-age screening.
  • Refining ICD labels into severity scores should strengthen the residual-disease association, since the paper's binary labels mix well-managed and poorly managed cases.
  • Any future BARB study should stratify by age before testing residual-disease relationships, since age dominates both ICD count and residual behavior.
  • The model's test-set $R^2$ of 0.816 with MAE of 5.45 years provides a realistic performance benchmark for small, single-site brain-age models.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial extension: the headline association is most fragile where the paper's own evidence is weakest—the age-unadjusted ANOVA on residuals grouped by number of ICD codes was null ($p = 0.1880$), so the positive result depends on the over-49 subgroup and on regression-line comparisons that may absorb age-composition effects.
  • If the BARB is real, it would most naturally be tested against disease severity (A1C, blood pressure readings, toxicology-confirmed substance use) rather than binary ICD codes, which the paper itself identifies as the main labeling limitation.
  • The use of routine T2-weighted clinical slices, if reproducible, would make the biomarker cheap enough to screen retrospectively on existing scans, without dedicated research acquisitions.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper develops a brain-age prediction model for 1,220 U.S. veterans using four CNNs trained on 2D T2-weighted FSE and FLAIR MRI slices at two anatomical locations, combined through a degree-3 polynomial ensemble. The model achieves R²=0.816 on the test set with honest reporting of training-set overfitting (training R²=0.97). The central claim of the paper is that the model's residuals constitute a biomarker for latent health conditions, based on an analysis where residuals grouped by the number of ICD-coded conditions (HTN, DM, mTBI, SAD, AAD) reportedly show statistically significant different trends (p=0.002), particularly for subjects older than 49 years.

Significance. If the biomarker claim held, the paper would add a modest but useful data point to the brain-age residual literature, especially because it uses T2-weighted images rather than the dominant T1 modality and reports overfitting explicitly. The age-prediction model itself is competently built and its performance is reported with commendable transparency, including cross-validation metrics, test-set metrics, and a comparison of training and test R². However, the central biomarker claim is not supported by the evidence presented. The headline p=0.002 comes from a test that is confounded by age composition across ICD-count groups, and the paper's own direct one-way ANOVA on residuals grouped by ICD count is null (p=0.1880). The over-49 subgroup result is post-hoc and uncorrected. The manuscript therefore does not establish the load-bearing claim that the residual is a usable biomarker.

major comments (3)
  1. [Abstract; §3.2 (Figure 7)] The headline claim that residuals grouped by the number of ICD codes show statistically significant different trends (p=0.002) is based on an F-test comparing predicted-age-versus-actual-age regression lines across ICD-count groups. This test does not directly compare residuals; it can reject the null when groups have different age distributions even if the residual distributions are identical. Because brain-age models typically exhibit regression to the mean, residuals are correlated with chronological age, so any two groups with different age compositions will show different regression lines. The paper itself notes in §3.2 that decision trees consistently split at actual age 48–49 and that there is a sharp decline in the proportion of subjects with zero ICD codes after this threshold, confirming that ICD-count groups differ in age composition. The direct one-way ANOVA on residuals grouped by ICD count, reported later in the same section, is null (p=0.1880). The p=0.002 result is therefore most plausibly an artifact of age confounding rather than evidence of an association between residuals and disease burden.
  2. [§3.2 (over-49 subgroup analysis)] The significant result restricted to subjects older than 49 (p=0.0069) is post-hoc and cannot rescue the null aggregate ANOVA. The threshold was selected after exploratory decision-tree analysis on the same data, and no multiple-comparison correction is reported. Moreover, the groups with 3–4 ICD codes in this age range have small sample sizes, as reflected in Table 6 and Figure 8, making the subgroup test unreliable. A post-hoc subgroup analysis without correction does not provide support for the claim that negative residuals correlate with multiple ICD codes in older subjects.
  3. [§3.2 (Table 6)] Even the descriptive statistics in Table 6 do not support the paper's narrative of a monotonic trend toward older predicted brain age with more ICD codes. The mean residuals for 0, 1, 2, 3, and 4 codes are +0.07, +0.21, +0.12, −0.82, and −0.72, respectively. The 1-code group has a more positive mean residual than the 0-code group, and the 2-code group is nearly identical to the 0-code group. Only the small 3- and 4-code groups show negative means. Combined with the null one-way ANOVA, the numeric pattern in Table 6 does not substantiate the abstract's assertion of a clear relationship between multiple ICD codes and advanced brain aging.
minor comments (6)
  1. [§1.3] The text says 'these conditions would benefit from a BADB'; this should read 'BARB'.
  2. [Figure 4] The figure contains a typo: 'Batch Normzalizatoin' should be 'Batch Normalization'.
  3. [§1.3] The phrase 'ICD codes can lack the subtly of distinguishing' should read 'subtlety'.
  4. [§3.2] The reported p-values of 'p = 0.000' for the HTN and SAD ANOVAs should be reported as 'p < 0.001' rather than an exact zero.
  5. [§4] The sentence 'The development for a clinically applicable BARB rests on on the need for...' contains a duplicated 'on'.
  6. [§2.1] The manuscript does not state whether institutional review board approval or patient consent was obtained for the MRI and ICD data; this should be clarified for a human-subjects study.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the residual-biomarker claim is not equivalent to its inputs, though it is statistically fragile.

full rationale

The derivation chain is not circular. The age model is trained on MRI slices with chronological age as labels; the five ICD codes are external chart-derived labels that never enter the CNN or ensemble fitting. The ensemble's reported R2=0.816 is on a held-out 221-subject test set, and Table 2 benchmarks it against independent studies, so the prediction claim is self-contained. The residual analysis uses residuals defined as actual minus predicted age, and tests their association with ICD-code count; because ICD count was not an input to the fitted model, a significant residual-group association would be an independent empirical finding, not a fit renamed as prediction. The paper is transparent that the direct one-way ANOVA on residuals grouped by ICD count was null ('A one-way ANOVA test of residuals grouped by the number of ICD codes did not yield statistically significant differences across the groups (p = 0.1880)'), and the aggregate trend-line test (F=4.265, p=0.002) plus the over-49 split (p=0.0069) are correctly viewed as potentially confounded by age composition and post-hoc threshold selection. That is a statistical correctness concern, not circularity: there is no equation in the paper that equals its own input by construction. Reference [8] is authored by overlapping team members, but it is not cited anywhere in the derivation of the ensemble or the residual analysis, so the self-citation is not load-bearing. No circular step can be exhibited.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

No new physical entities are introduced; BARB is a naming convention for an existing residual biomarker concept. The model relies on standard CNN components and polynomial regression, with the listed free parameters and assumptions. The central claim rests on the age-prediction model, the residual biomarker assumption, and the post hoc subgroup analysis.

free parameters (3)
  • Ensemble polynomial coefficients (a_i, b_i, c_i, d) = Not reported
    The 15 coefficients in Equation 2 are fitted to training-set CNN predictions to maximize age accuracy. Their values are not listed, so the residual analysis cannot be reproduced or checked for stability.
  • Age split threshold for subgroup analysis = 49 years
    Chosen from decision-tree splits on the same data (Section 3.2, split near ages 48-49) and then used as the cutoff for the significant ANOVA (p=0.0069). The threshold is post hoc, which inflates the reported significance.
  • CNN training hyperparameters = See Figure 4
    Filter counts (16, 32, 64), dense width (16), dropout (0.5), batch size (20), and pooling size (2x2) are reported, but optimizer, learning rate, number of epochs, and random seed are not. These choices affect the residuals and are not varied in a sensitivity analysis.
assumptions (4)
  • standard math The squared prediction error decomposes into bias squared, variance, and irreducible error (Eq. 1).
    Used in Section 1.1 to justify why residuals might contain disease signal; standard result from Hastie et al. [51].
  • domain assumption The irreducible error is dominated by external health factors, so residual signals can be read as disease biomarkers.
    Section 1.1 states the irreducible error 'must be an outsized portion of the residual and primarily influenced by the target health factor(s)'; this is asserted, not demonstrated, and underpins the residual analysis.
  • domain assumption ICD-code mislabeling occurs randomly and does not systematically bias the residual-disease relationship.
    Section 3.2 assumes 'labeling inconsistencies and misclassification are assumed to occur randomly across the dataset'; if mislabeling is correlated with age or severity, the residual trends could be biased.
  • domain assumption Having more ICD codes corresponds on average to more advanced brain aging.
    Section 2.1 states 'an older appearing brain is more likely to have at least one of these ICD codes.' This is the hypothesis under test, not an independently established premise, yet the count-based grouping treats it as meaningful.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Brain Age Residual Biomarker (BARB): Leveraging MRI-Based Models to Detect Latent Health Conditions in U.S. Veterans." pith.science (2026). https://pith.science/paper/46YMRLJI

@misc{pith2026250105970,
  author       = {Pith},
  title        = {Pith review of: A Brain Age Residual Biomarker (BARB): Leveraging MRI-Based Models to Detect Latent Health Conditions in U.S. Veterans},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/46YMRLJI}},
  note         = {Machine review of arXiv:2501.05970}
}
abstract

Age prediction using brain imaging, such as MRIs, has achieved promising results, with several studies identifying the model's residual as a potential biomarker for chronic disease states. In this study, we developed a brain age predictive model using a dataset of 1,220 U.S. veterans (18--80 years) and convolutional neural networks (CNNs) trained on two-dimensional slices of axial T2-weighted fast spin-echo and T2-weighted fluid attenuated inversion recovery MRI images. The model, incorporating a degree-3 polynomial ensemble, achieved an $R^{2}$ of 0.816 on the testing set. Images were acquired at the level of the anterior commissure and the frontal horns of the lateral ventricles. Residual analysis was performed to assess its potential as a biomarker for five ICD-coded conditions: hypertension (HTN), diabetes mellitus (DM), mild traumatic brain injury (mTBI), illicit substance abuse/dependence (SAD), and alcohol abuse/dependence (AAD). Residuals grouped by the number of ICD-coded conditions demonstrated different trends that were statistically significant ($p = 0.002$), suggesting a relationship between disease states and predicted brain age. This association was particularly pronounced in patients over 49 years, where negative residuals (indicating advanced brain aging) correlated with the presence of multiple ICD codes. These findings support the potential of residuals as biomarkers for detecting latent health conditions.

Figures

Figures reproduced from arXiv: 2501.05970 by the authors.

Figure 1
Figure 1. The model presented here is constructed from T2 weighted fast spin-echo (FSE) and T2 weighted fluid attenuated inversion recovery (FLAIR) images of two locations of the brain: the anterior commissure and the frontal horns of the lateral ven￾tricles. This study contributes to the growing field of brain age resid￾ual biomarkers (BARBs) by uti￾lizing a unique dataset of 1,220 U.S. veterans aged 20–80. Each participant … view at source ↗
Figure 2
Figure 2. Distribution of the ICD codes, with a total of 1220 patients. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Histogram illustrating the distribution of contin [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Architecture of the convolutional neural network (CNN) model used for each image-type. [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Four types of ensemble methods were evaluated for this study, as outlined in Table 3. [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Predicted age vs. real age for train and test sets, color-coded [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Predicted age vs. real age for train and test sets, color [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Histograms of residuals broken down by ICD codes (see [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 40 canonical work pages

  1. [1]

    W., Bhowmik, A

    An, W. W., Bhowmik, A. C., Nelson, C. A., & Wilkinson, C. L. (2024). EEG-based brain age prediction in infants-toddlers: Implications for early detection of neurodevelopmental disorders. Developmental Cognitive Neuroscience, 101493.https://doi.org/10.1016/j.dcn.2024.101493

    work page arXiv 2024

  2. [2]

    G., Maximov, I

    Beck, D., de Lange, A.-M. G., Maximov, I. I., Richard, G., Andreassen, O. A., Nordvik, J. E., & Westlye, L. T. (2021). White matter microstructure across the adult lifespan: A mixed longitudinal and cross-sectional study using advanced diffusion models and brain-age prediction. NeuroImage, 224, 117441.https://doi.org/10.1016/j.neuroimage.2020.117441

    work page arXiv 2021

  3. [3]

    Beheshti, I., Nugent, S., Potvin, O., & Duchesne, S. (2019). Bias-adjustment in neuroimaging- based brain age frameworks: A robust scheme.NeuroImage: Clinical, 24, 102063.https://doi. org/10.1016/j.nicl.2019.102063

    arXiv 2019

  4. [4]

    M., Rizzo, G

    Boyle, R., Jollans, L., Rueda-Delgado, L. M., Rizzo, G. G., Yener, G. G., McMorrow, J. P., Knight, S. P., & Carey, D. (2021). Brain-predicted age difference score is related to specific cognitive functions: A multi-site replication analysis.Brain Imaging and Behavior, 15(1), 327–

    work page 2021

  5. [5]

    Chen, C.-L., Hsu, Y.-C., Yang, L.-Y., Tung, Y.-H., Luo, W.-B., Liu, C.-M., & Tseng, W.-Y. I. (2020). Generalization of diffusion magnetic resonance imaging-based brain age prediction model through transfer learning.NeuroImage, 217, 116831.https://doi.org/10.1016/j.neuroimage. 2020.116831

    arXiv 2020

  6. [6]

    Cho, H.-H., Kim, J., Na, I., Song, H.-N., Choi, J.-U., Baek, I.-Y., Lee, J.-E., Chung, J.-W., Kim, C.-K., Oh, K., Bang, O.-Y., Kim, G.-M., Seo, W.-K., & Park, H. (2024). Predicting cerebrovas- cular age and its clinical relevance: Modeling using 3D morphological features of brain vessels. Heliyon, 10, e32375.https://doi.org/10.1016/j.heliyon.2024.e32375

    work page doi:10.1016/j.heliyon.2024.e32375 2024

  7. [7]

    H., Ritchie, S

    Cole, J. H., Ritchie, S. J., Bastin, M. E., Valdés Hernández, M. C., Maniega, S. M., Royle, N., Corley, J., Starr, J. M., Harris, S. E., Wray, N. R., Redmond, P., Marioni, R. E., Wardlaw, J. 17 M., Sharp, D. J., Deary, I. J., Cox, S. R., & Ward, J. M. (2018). Brain age predicts mortality. Molecular Psychiatry, 23, 1385–1392.https://doi.org/10.1038/mp.2017.62

    work page doi:10.1038/mp.2017.62 2018

  8. [8]

    Conneely, M., Swaminathan, B., Vyas, P., Bousquet, A., Jamshidi Zelenberg, S., Banerji, S., Piao, X., Fatima, A., Patel, P., Burke, J., Golant, B., Nolcheff, R. (2022). Evaluating the Clinical Relevance of Age Prediction/Actual Age Discrepancy utilizing a convolutional neural network paired with a regression algorithm applied to T2 weighted fast spin-echo...

    work page 2022

Show all 55 references
  1. [9]

    B., Westman, E., & AddNeuroMed Consortium (2024)

    Dartora, C., Marseglia, A., Mårtensson, G., Rukh, G., Dang, J., Muehlboeck, J.-S., Wahlund, L.-O., Moreno, R., Barroso, J., Ferreira, D., Schiöth, H. B., Westman, E., & AddNeuroMed Consortium (2024). A deep learning model for brain age prediction using minimally preprocessed T...

    2024

  2. [10]

    Dular, L., Pernuš, F., & Špiclin, Ž. (2024). Extensive T1-weighted MRI preprocessing improves generalizability of deep brain age prediction models.Computers in Biology and Medicine, 173, 108320. https://doi.org/10.1016/j.compbiomed.2024.108320

    2024

  3. [11]

    Fang, Z., Pan, N., Liu, S., Li, H., Pan, M., Zhang, J., Li, Z., Liu, M., & Ge, X. (2024). Comparative analysis of brain age prediction using structural and diffusion MRIs in neonates. NeuroImage, 299, 120815.https://doi.org/10.1016/j.neuroimage.2024.120815

    2024

  4. [12]

    Franke, K., Ziegler, G., Klöppel, S., & Gaser, C. (2010). Estimating the age of healthy subjects fromT1-weightedMRIscansusingkernelmethods: Exploringtheinfluenceofvariousparameters. NeuroImage, 50, 883–892.https://doi.org/10.1016/j.neuroimage.2010.01.005

    2010 doi

  5. [13]

    A., Haque, F., Roy, T., Islam, M., Nahiduzzaman, M

    Hasan, M. A., Haque, F., Roy, T., Islam, M., Nahiduzzaman, M. M., Hasan, M. M., Ahsan, M., & Haider, J. (2024). Prediction of fetal brain gestational age using multihead attention with Xception.Computers in Biology and Medicine, 182, 109155.https://doi.org/10.1016/j. compbiome...

    2024

  6. [14]

    K., Lee, J

    Hwang, I., Yeon, E. K., Lee, J. Y., Yoo, R.-E., Kang, K. M., Yun, T. J., Choi, S. H., Sohn, C.-H., Kim, H., & Kim, J.-H. (2021). Prediction of brain age from routine T2-weighted spin- echo brain magnetic resonance images with a deep convolutional neural network.Neurobiology of...

    2021 doi

  7. [15]

    A., Bjornsdottir, G., Thorgeirsson, T

    Jonsson, B. A., Bjornsdottir, G., Thorgeirsson, T. E., Ellingsen, L. M., Gudbjartsson, D. F., Walters, G. B., Deary, I. J., Stefansson, H., Stefansson, K., Ulfarsson, M. O. (2019). Brain age prediction using deep learning uncovers associated sequence variants.Nature Communicat...

    2019 doi

  8. [16]

    Kianian, I., & Sajedi, H. (2024). Brain age estimation with a greedy dual-stream model for lim- ited datasets.Neurocomputing, 596, 127974.https://doi.org/10.1016/j.neucom.2024.127974

    2024

  9. [17]

    T., & Maximov, I

    Korbmacher, M., Westlye, L. T., & Maximov, I. I. (2024). FreeSurfer version-shuffling can enhance brain age predictions.NeuroImage: Reports, 4, 100214. https://doi.org/10.1016/j. ynirp.2024.100214 18

    2024

  10. [18]

    Li, X., Hao, Z., Li, D., Jin, Q., Tang, Z., Yao, X., & Wu, T. (2024). Brain age prediction via cross-stratified ensemble learning.NeuroImage, 299, 120825. https://doi.org/10.1016/j. neuroimage.2024.120825

    2024

  11. [19]

    K., Huntenburg, J

    Liem, F., Varoquaux, G., Kynast, J., Beyer, F., Masouleh, S. K., Huntenburg, J. M., Lampe, L., Rahim, M., Abraham, A., Craddock, R. C., Riedel-Heller, S., Luck, T., Loeffler, M., Schroeter, M. L., Witte, A. V., Villringer, A., & Margulies, D. S. (2017). Predicting brain-age fr...

    2017

  12. [20]

    K., & Sundarrajan, R

    Soumya Kumari, L. K., & Sundarrajan, R. (2024). A review on brain age prediction models. Brain Research, 1823, 148668.https://doi.org/10.1016/j.brainres.2023.148668

    2024

  13. [21]

    The effect of head motion on brain age prediction using deep convolutional neural networks

    Vakli, P., Weiss, B., Rozmann, D., Erőss, G., Nárai, Á., Hermann, P., &Vidnyánszky, Z.(2024). The effect of head motion on brain age prediction using deep convolutional neural networks. NeuroImage, 294, 120646.https://doi.org/10.1016/j.neuroimage.2024.120646

    2024

  14. [22]

    A., Laffitte Nodarse, C., Cole, J

    Valdes-Hernandez, P. A., Laffitte Nodarse, C., Cole, J. H., & Cruz-Almeida, Y. (2023). Fea- sibility of brain age predictions from clinical T1-weighted MRIs.Brain Research Bulletin, 205, 110811. https://doi.org/10.1016/j.brainresbull.2023.110811

    2023

  15. [23]

    Zhang, X., Pan, Y., Wu, T., Zhao, W., Zhang, H., Ding, J., Ji, Q., Jia, X., Li, X., Lee, Z., Zhang, J., & Bai, L. (2024). Brain age prediction using interpretable multi-feature-based convolutional neural network in mild traumatic brain injury.NeuroImage, 297, 120751. https: //...

    2024

  16. [24]

    R., & Bataller, R

    Cabezas, J., Lucey, M. R., & Bataller, R. (2016). Biomarkers for Monitoring Alcohol Use. Clinical Liver Disease, 8(3), 59–63.https://doi.org/10.1002/cld.571

    2016 doi

  17. [25]

    Desai, A., Chen, H., & Kim, H.-Y. (2020). Multiple Mild Traumatic Brain Injuries Lead to Visual Dysfunction in a Mouse Model.Journal of Neurotrauma, 37, 286–294.https://doi.org/ 10.1089/neu.2019.6602

    2020

  18. [26]

    Gąsecki, D., Kwarciany, M., Nyka, W., & Narkiewicz, K. (2013). Hypertension, Brain Damage and Cognitive Decline.Current Hypertension Reports, 15, 547–558.https://doi.org/10.1007/ s11906-013-0398-4

    2013

  19. [27]

    Gonzalo, P., Radenne, S., & Gonzalo, S. (2014). Biomarkers of chronic alcohol misuse.Current Biomarker Findings, 4, 9–22.https://doi.org/10.2147/CBF.S37239

    2014 doi

  20. [28]

    Lingford-Hughes, A. (2005). Human brain imaging and substance abuse.Current Opinion in Pharmacology, 5(1), 42–46.https://doi.org/10.1016/j.coph.2004.10.002

    2005 doi

  21. [29]

    Moheet, A., Mangia, S., & Seaquist, E. R. (2015). Impact of diabetes on cognitive function and brain structure.Annals of the New York Academy of Sciences, 1353, 60–71. https://doi.org/ 10.1111/nyas.12807

    2015 doi

  22. [30]

    Oscar-Berman, M., & Marinkovic, K. (2003). Alcoholism and the brain: an overview. Al- cohol Research and Health, 27(2), 125–133. https://www.ncbi.nlm.nih.gov/pmc/articles/ PMC6668884/ 19

    2003

  23. [31]

    Peterson, K. (2004). Biomarkers for Alcohol Use and Abuse: A Summary.Alcohol Research and Health, 28(1), 30–37.https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6601655/

    2004

  24. [32]

    (n.d.).Alzheimer’s Disease Neuroimaging Initia- tive (ADNI)

    Alzheimer’s Disease Neuroimaging Initiative. (n.d.).Alzheimer’s Disease Neuroimaging Initia- tive (ADNI). https://adni.loni.usc.edu/

  25. [33]

    (n.d.).ICBM Dataset

    International Consortium for Brain Mapping (ICBM). (n.d.).ICBM Dataset. https://ida. loni.usc.edu/collaboration/access/appLicense.jsp

  26. [34]

    Japanese Alzheimer’s Disease Neuroimaging Initiative (J-ADNI). (n.d.). J-ADNI Dataset. https://www.gaaindata.org/partner/J-ADNI

  27. [35]

    Biomedical Image Analysis Group. (n.d.). IXI Dataset. https://brain-development.org/ ixi-dataset/

  28. [36]

    University of Florida Health. (n.d.). UFHealth Dataset . https://idr.ufhealth.org/ research-services/research-data/

  29. [37]

    (n.d.).UK Biobank

    UK Biobank Limited. (n.d.).UK Biobank. https://www.ukbiobank.ac.uk

  30. [38]

    A., Bush, A

    Ellis, K. A., Bush, A. I., Darby, D., De Fazio, D., Foster, J., Hudson, P., Lautenschlager, N. T., Lenzo, N., Martins, R. N., Maruff, P., Masters, C., Milner, A., Pike, K., Rowe, C., Savage, G., Szoeke, C., Taddei, K., Villemagne, V., Woodward, M., Ames, D., & AIBL Re- search ...

    2009

  31. [39]

    J., Gow, A

    Deary, I. J., Gow, A. J., Pattie, A., & Starr, J. M. (2012). Cohort profile: The Lothian Birth Cohorts of 1921 and 1936.International Journal of Epidemiology, 41, 1576–1584. https: //doi.org/10.1093/ije/dyr197

    2012 doi

  32. [40]

    J., Gow, A

    Deary, I. J., Gow, A. J., Taylor, M. D., Corley, J., Brett, C., Wilson, V., & others. (2007). The Lothian Birth Cohort 1936: A study to examine influences on cognitive ageing from age 11 to age 70 and beyond.BMC Geriatrics, 7, 28.https://doi.org/10.1186/1471-2318-7-28

    2007 doi

  33. [41]

    Ge, T., Chen, X., Li, M., Zheng, Y., Zhang, J., Liu, J., Liu, S., Zhou, Y., Chen, W., & Qiu, J. (2023). The Chinese Human Connectome Project: A Multimodal MRI Dataset for Brain Research. Nature Neuroscience, 26, 78–86.https://doi.org/10.1038/s41593-022-01121-3

    2023 doi

  34. [42]

    M., Rodrigue, K

    Kennedy, K. M., Rodrigue, K. M., Bischof, G. N., Hebrank, A. C., Reuter-Lorenz, P. A., & Park, D. C. (2015). Age trajectories of functional activation during associative memory encoding: A cross-sectional and longitudinal investigation.Neurobiology of Aging, 36(4), 1490–1503.h...

    2015 doi

  35. [43]

    Loeffler, M., Engel, C., Ahnert, P., Alfermann, D., Arelin, K., Baber, R., Beutner, F., Binder, H., Brähler, E., Burkhardt, R., Ceglarek, U., Enzenbach, C., Fuchs, M., Glaesmer, H., Girlich, F., Hagendorff, A., Häntzsch, M., Hegerl, U., Henger, S., Hensch, T., Hinz, A., Holzen...

    2015

  36. [44]

    Lovestone, S., Francis, P., Kloszewska, I., Mecocci, P., Simmons, A., Soininen, H., Spenger, C., Tsolaki, M., Vellas, B., Wahlund, L.-O., Ward, M., & AddNeuroMed Consortium. (2009). AddNeuroMed–the European collaboration for the discovery of novel biomarkers for Alzheimer’s di...

    2009

  37. [45]

    B., Colcombe, S

    Nooner, K. B., Colcombe, S. J., Tobe, R. H., Mennes, M., Benedict, M., Moreno, A., Panek, L., Brown, S., Zavitz, S., Li, Q., Sikka, S., Gutman, D., Bangaru, S., Schlachter, R., Kamiel, S. M., Anwar, A. R., Hinz, C. M., Kaplan, M. S., Rachlin, A. B., Adelsberg, S., Cheung, B., ...

    2012 arXiv

  38. [46]

    A., Tyler, L

    Shafto, M. A., Tyler, L. K., Dixon, M., Taylor, J. R., Rowe, J. B., Cusack, R., Calder, A. J., Marslen-Wilson, W. D., Duncan, J., Dalgleish, T., Henson, R. N. (2014). The Cambridge Centre for Ageing and Neuroscience (Cam-CAN): A Cross-Sectional, Multi-Modal Imaging Dataset for...

    2014 doi

  39. [47]

    R., Williams, N., Cusack, R., Auer, T., Shafto, M

    Taylor, J. R., Williams, N., Cusack, R., Auer, T., Shafto, M. A., Tyler, L. K., Calder, A. J., & Henson, R. N. (2017). The Cambridge Centre for Ageing and Neuroscience (Cam-CAN): Understanding Variability in Healthy Ageing Using a Cross-Sectional, Multi-Modal Approach. Nature ...

    2017 doi

  40. [48]

    H., Nooner, K

    Tobe, R. H., Nooner, K. B., Keator, D., Adelsberg, S., Rokem, A., & Milham, M. P. (2022). The NKI-Rockland Sample: Advancing the Open Science Movement in Psychiatry and Neuroscience. Frontiers in Psychiatry, 13, 821415.https://doi.org/10.3389/fpsyt.2022.821415

    2022

  41. [49]

    Wei, D., Yang, J., Li, W., Wang, K., Zhang, Q., & Qiu, J. (2018). ALFF, ReHo and Functional Connectivity of Resting-State fMRI in Non-Clinical Depressive Symptoms: A Study Using the Southwest University Adult Lifespan Dataset (SALD).Frontiers in Human Neuroscience, 12, 107. ht...

    2018

  42. [50]

    S., Bellec, P., Birn, R

    Zuo, X.-N., Anderson, J. S., Bellec, P., Birn, R. M., Biswal, B. B., Blautzik, J., Breitner, J. C. S., Buckner, R. L., Calhoun, V. D., Castellanos, F. X., Chen, A., Chen, B., Chen, G., Colcombe, S. J., Courtney, S., Dagher, A., Davatzikos, C., Di Martino, A., Dosenbach, N. U. ...

    2014

  43. [51]

    Hastie, T., Tibshirani, R., &Friedman, J.(2001).The Elements of Statistical Learning.Springer New York Inc. 21

    2001

  44. [52]

    Patterns, 2(8), 100311

    Gao, J.(2021).Bias-variancedecompositionforthemeanabsoluteerror. Patterns, 2(8), 100311. https://doi.org/10.1016/j.patter.2021.100311

    2021

  45. [53]

    Goceri, E. (2023). Medical image data augmentation: techniques, comparisons and in- terpretations. Artificial Intelligence Review , 56, 12561–12605. https://doi.org/10.1007/ s10462-023-10280-y

    2023

  46. [54]

    Joshi, P., & Séquin, C. (2010). An intuitive explanation of third-order surface behavior.Com- puter Aided Geometric Design, 27, 150–161.https://doi.org/10.1016/j.cagd.2009.11.003 22

    2010 doi

  47. [345]

    https://doi.org/10.1007/s11682-020-00260-3

    doi

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.