Pith. sign in

REVIEW 1 major objections 9 minor 46 references

AT-Attn: Temporal-Aware Cross-Attention for Longitudinal Multimodal Alzheimer's Disease Diagnosis

T0 review · 1 major / 9 minor · reviewed 2026-07-09 · glm-5.2

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

Pith's one-line read One-way cross-attention lets noisy brain scans help—not hurt—AD diagnosis

desk verdict Solid architecture for constrained multimodal fusion, but the 'MRI adds complementary information' claim is confounded by a gap between AT-Attn's clinical branch and XGB. read the letter →

arxiv 2607.07091 v1 pith:NUGQF7LT submitted 2026-07-08 cs.CV cs.AI

Xinyue Du , Yibo Liu , Zhenglei Zhou , Xuancheng Yao , Weimin Zhong , Qiuhui Chen This is my paper · ORCID
classification cs.CVcs.AI
keywords Alzheimer'sdiseasediagnosismultimodalfusioncross-attentionlongitudinalmodelingstructuralMRIcognitiveassessmentirregulartimeseriesasymmetricattention
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

The paper argues that in longitudinal Alzheimer's disease diagnosis, the problem is not whether to combine brain MRI with cognitive test scores, but how. MRI is high-dimensional, noisy, and often missing at some visits; cognitive scales are more directly diagnostic. When you fuse them naively—by concatenation or symmetric cross-attention—the weaker MRI signal can distort the stronger clinical signal and degrade performance. The authors propose a constrained fusion architecture called AT-Attn built around three ideas: a Change-and-Time encoder that captures how both MRI features and cognitive scores change between visits without computing unstable temporal derivatives; an asymmetric cross-attention layer where MRI features query cognitive-scale features but not the reverse, so clinical context refines MRI representations without letting noisy MRI feedback contaminate the clinical branch; and a learnable time bias that down-weights temporally distant visits in the attention computation. A gated fusion step then blends the cross-attention-enhanced MRI with the original MRI features, and shortcut paths preserve direct unimodal evidence for the final classifier. On an ADNI cohort of 1,520 patients with patient-level five-fold cross-validation, this architecture outperforms unimodal MRI, unimodal cognitive scales, and all naive fusion baselines (late fusion, concatenation, modality dropout, missing-aware gated fusion) across accuracy, macro F1, ROC-AUC, and PR-AUC. It remains competitive with strong tabular baselines like XGBoost and LightGBM. Subgroup analysis shows MRI contribution is context-dependent: gains are largest when baseline impairment is already substantial, under moderate MRI missingness, and with longer follow-up—consistent with the clinical understanding that structural MRI becomes more informative as neurodegeneration progresses.

What carries the argument

AT-Attn combines: (1) a Change-and-Time encoder computing inter-visit feature differences with normalized elapsed time, absolute temporal position, and instantaneous state, using separate parameters per modality; (2) time-biased asymmetric cross-attention where MRI features serve as queries and cognitive-scale features as keys/values, with an additive penalty of the form −exp(|Δt|/τ_h) per attention head on pre-softmax logits, where τ_h is learnable and positive; (3) gated fusion blending raw and cross-attended MRI features via a per-timestep sigmoid gate; (4) shortcut paths exposing direct unimodal pooled features alongside fused features to the classifier; (5) a 3D ResNet-18 MRI backbone (

What would settle it

If a variant that excludes the cognitive scales most directly used in ADNI diagnostic criteria (e.g., ADAS11, ADAS13) from the input features shows no MRI contribution gain over clinical-only baselines, then the MRI value demonstrated here depends on label leakage rather than independent structural information.

Watch

Extended reading notes

Core claim

The central claim is that structural MRI provides complementary diagnostic value in longitudinal AD staging only when integrated through a constrained, temporally-aware fusion mechanism—not when added naively. The load-bearing mechanism is the asymmetric direction of cross-attention: MRI queries cognitive scales, not the reverse, which prevents the weaker modality from corrupting the stronger one while still allowing cross-modal enrichment. Combined with temporal change encoding and a learnable exponential time penalty on attention logits, this architecture extracts clinically relevant MRI signal that naive fusion cannot.

Load-bearing premise

The paper assumes that ADNI clinical diagnoses used as training labels are not largely a deterministic function of the cognitive-scale features used as inputs. The authors acknowledge this risk themselves: if diagnoses are substantially defined by the same cognitive tests fed to the model, then the model may be fitting label leakage rather than learning independent biomarker value from MRI.

Editorial extensions

If this is right

  • If the asymmetric fusion principle generalizes, any multimodal clinical setting with unequal modality strength—where one signal is noisier or less directly diagnostic—could benefit from one-way cross-attention rather than symmetric fusion, including PET-MRI, EEG-clinical, or pathology-radiology combinations.
  • The learnable temporal bias on attention logits offers a general mechanism for irregularly-sampled multimodal data, potentially applicable beyond AD to any longitudinal disease staging where visit spacing varies across patients.
  • The finding that MRI contribution is context-dependent (larger in already-impaired patients, under moderate missingness, with longer follow-up) suggests that multimodal AI systems should report subgroup-level value rather than aggregate metrics alone, as a modality that helps on average may hurt specific patient subgroups.
  • The shortcut-path design—preserving direct unimodal evidence alongside fused features—implies that fusion architectures should never fully replace a modality's raw representation, only augment it, when that modality carries independent diagnostic signal.

Reading between the lines

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

  • The asymmetric principle may extend to a general design rule: in any multimodal system with a dominant and a complementary modality, cross-attention should flow from dominant to complementary, never bidirectionally, unless both modalities are comparably reliable. This is not tested in the paper but is a natural extrapolation from the ablation showing symmetric cross-attention performs worse on acc
  • The context-dependent MRI contribution pattern (largest gains in DEM subgroup) raises the possibility that the architecture is effectively learning to weight MRI more when cognitive scores are already abnormal—which is what clinical intuition would prescribe. A system that explicitly conditions fusion strength on estimated impairment severity might achieve the same effect more transparently.
  • If the label-leakage concern the authors acknowledge is significant—ADNI diagnoses are partly determined by the same cognitive scales used as inputs—then the true independent biomarker value of MRI may be higher than reported, because the model's ceiling is artificially capped by labels that partially encode the input features.
  • The time-biased attention pattern (nearest visit accounts for only ~25% of attention peaks, distant visits ~50%) suggests the model discovers that diagnostically relevant MRI information is distributed across the trajectory rather than concentrated at the most recent scan, which has implications for how clinical visits should be scheduled and prioritized.
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

1 major / 9 minor

Summary. The paper proposes AT-Attn, a temporal-aware multimodal framework for longitudinal Alzheimer's disease (AD) diagnosis. The architecture combines (1) a Change-and-Time encoder that captures inter-visit feature differences and absolute temporal position, (2) time-biased asymmetric cross-attention in which MRI features query cognitive-scale features under an additive exponential temporal penalty, and (3) gated fusion with shortcut paths that preserve direct unimodal evidence. The framework is evaluated on an MRI-retained ADNI cohort of 1,520 patients with patient-level five-fold cross-validation, against unimodal baselines, naive fusion baselines, and strong tabular baselines (XGB, LGBM, etc.). The main asymmetric AT-Attn model achieves accuracy 0.719, macro F1 0.721, ROC-AUC 0.873, and PR-AUC 0.783, outperforming naive fusion baselines and remaining competitive with tree-based tabular methods. Systematic ablations (Table III), subgroup analyses (Fig. 3), attention pattern analyses (Fig. 4), and representative corrected cases (Table IV) are provided.

Significance. The problem of stable integration of a weaker modality (structural MRI) with a stronger clinical signal under irregular longitudinal sampling is well-motivated and practically relevant. The architectural design is principled: the asymmetric cross-attention direction (MRI queries cognitive scales, not vice versa) is a concrete and falsifiable design choice motivated by modality strength imbalance, and the ablation confirming that removing Change-and-Time encoding causes the strongest component-level degradation (Table III) supports the temporal modeling contribution. The subgroup gain analysis (Fig. 3) showing context-dependent MRI contribution—particularly larger gains when baseline diagnosis is DEM—is a clinically plausible and informative finding. The paper is transparent about the label-leakage limitation inherent to ADNI studies (§IV-F).

major comments (1)
  1. §IV-B, Table II vs. Table III: The central interpretive claim that MRI provides 'complementary information' beyond clinical features is confounded by the gap between the clinical-only ablation and the XGB baseline. Full AT-Attn with MRI achieves 0.7192 Acc (Table III), which is approximately equal to XGB without MRI (0.7187 Acc, Table II). Meanwhile, AT-Attn without MRI (clinical-only matched, 0.6935 Acc, Table III) is substantially worse than XGB without MRI (0.7187). This pattern is consistent with MRI compensating for AT-Attn's weaker clinical-feature processing relative to XGB, rather than reflecting genuine biomarker complementarity. The paper does not address this confound directly. A controlled comparison—e.g., adding MRI to the XGB clinical feature set, or demonstrating that AT-Attn's clinical branch matches XGB before MRI is added—would substantially strengthen the 'complementar
minor comments (9)
  1. §III-D, Eq. (8): The text states that '-exp(Δt/τ_h) explicitly down-weights temporally distant keys, whereas exp(-Δt/τ_h) mainly gives nearby visits a small positive bonus.' The two expressions are mathematically equivalent up to sign convention in the exponent; the distinction as stated is unclear. Consider rephrasing to clarify that the adopted form is a negative penalty added to pre-softmax logits.
  2. §IV-A: The cohort funnel (Fig. 2A) reports 4,813 visits with matched MRI and 1,753 without, but Table I reports 5,176 preprocessed MRI scans. The difference (363 scans) is not explained. Clarify whether these scans were excluded during visit-grid matching and why.
  3. Table II: The 'MCI Sens@90Spe / Achieved Spe' column header is ambiguous. Consider splitting into two columns or clarifying in the caption that the format is 'sensitivity / achieved specificity at a threshold targeting 90% specificity.'
  4. §IV-D, Fig. 3A: The subgroup sizes (n=598, 682, 240) sum to 1,520, but Fig. 3D uses different subgroup sizes (n=592, 570, 358) that also sum to 1,520. The discrepancy in CN/MCI/DEM counts between panels A and D is unexplained.
  5. §III-C, Eq. (2): The normalization constant 12.0 for Δt_norm is introduced without justification. A brief note that this corresponds to a 12-month reference interval would aid interpretation.
  6. §IV-F: The limitation that 'ADNI clinical diagnoses are strongly tied to cognitive and functional assessments' is acknowledged, but the paper could strengthen its framing by noting this is a known issue across ADNI-based ML studies and citing prior work that has attempted to mitigate it (e.g., by excluding diagnosis-defining scales).
  7. Fig. 1: The diagram is informative but dense. The 'Learnable Time Bias' and 'Time-Biased Cross-Attention' boxes appear disconnected from the main data flow arrows. Consider adding arrows to clarify the flow from the time bias module to the cross-attention computation.
  8. §IV-E, Fig. 4A: The y-axis label 'Mean Attention Mass' could be clarified as 'mean attention weight' to distinguish from the pre-softmax attention logits.
  9. References: Several MICCAI 2025 citations (e.g., [29], [30], [31]) appear with volume numbers but no DOI. For a journal submission, complete bibliographic information should be provided.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: AT-Attn's results are empirically validated against external baselines, not derived by construction.

full rationale

The paper proposes a neural architecture (AT-Attn) and evaluates it empirically on the ADNI cohort. The claimed results (e.g., accuracy 0.719) are measured on held-out test folds, not derived analytically from the inputs. The architectural components (Change-and-Time encoding, time-biased cross-attention, gated fusion) are design choices, not mathematical identities that force the output to equal the input. The label leakage concern (labels partly defined by cognitive scales used as features) is a methodological limitation acknowledged in §IV-F, but it is not a circularity in the derivation chain: the model does not claim to predict the labels from a function that is defined as the labels. The comparison to XGB is an empirical benchmark, not a self-citation loop. No equations reduce to their inputs by construction. The derivation is self-contained and externally falsifiable via the cross-validation protocol. Score 0 is warranted.

Assumptions & free parameters 5 free parameters · 4 assumptions · 2 invented entities

The model relies on standard neural network parameters and domain-specific assumptions about modality strength. The exponential temporal bias is an ad-hoc choice validated empirically. No new physical entities or forces are postulated.

free parameters (5)
  • τ_h (per-head time bias scale) = learned via softplus
    Fitted per head to learn temporal decay rate
  • D (feature dimension) = 256
    Chosen architecture hyperparameter
  • H (attention heads) = 4
    Chosen architecture hyperparameter
  • Learning rates = 1e-4 (CNN), 1e-3 (rest)
    Standard hyperparameter choices
  • Gate MLP weights = trained
    Standard learned parameters
assumptions (4)
  • domain assumption Cognitive scales are more directly diagnostic than MRI in this cohort
    Motivates the asymmetric cross-attention direction (§III-D)
  • domain assumption ADNI clinical diagnoses represent valid ground truth for disease stage
    Required for supervised training
  • ad hoc to paper Exponential decay is an appropriate functional form for temporal bias
    Chosen for the time bias function in Eq. 8
  • domain assumption Patient-level label at last visit is the correct prediction target
    Defines the task formulation (§III-A)
invented entities (2)
  • Change-and-Time Encoder
    purpose: Encode temporal dynamics and feature changes
    Architectural module, validated by ablation
  • Time-Biased Asymmetric Cross-Attention
    purpose: Refine MRI with cognitive context under temporal penalty
    Architectural module, validated by ablation

how reviews work

0 comments
Cite this review

Pith. "Pith review of AT-Attn: Temporal-Aware Cross-Attention for Longitudinal Multimodal Alzheimer's Disease Diagnosis." pith.science (2026). https://pith.science/paper/NUGQF7LT

@misc{pith2026260707091,
  author       = {Pith},
  title        = {Pith review of: AT-Attn: Temporal-Aware Cross-Attention for Longitudinal Multimodal Alzheimer's Disease Diagnosis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NUGQF7LT}},
  note         = {Machine review of arXiv:2607.07091}
}
read the original abstract

In longitudinal Alzheimer's disease (AD) diagnosis support, clinical and imaging information is often collected at irregular visits. Integrating these multimodal observations may improve diagnostic assessment, but naive fusion can degrade performance when MRI is noisy or intermittently unavailable. We propose AT-Attn, a temporal-aware multimodal framework that combines Change-and-Time encoding, time-biased asymmetric cross-attention, and gated fusion to integrate MRI with longitudinal clinical information. We evaluate AT-Attn on an MRI-retained ADNI cohort of 1,520 patients using structural MRI, six cognitive-scale trajectories, and seven static clinical variables under patient-level five-fold cross-validation. The main asymmetric AT-Attn model achieves accuracy 0.719+/-0.024, macro F1 0.721+/-0.023, ROC-AUC 0.873+/-0.013, and PR-AUC 0.783+/-0.018, outperforming unimodal and naive multimodal fusion baselines while remaining competitive with strong tabular baselines. These results suggest that a temporal-aware and constrained fusion strategy can help structural MRI contribute clinically relevant complementary information for patient-level AD diagnosis support.

Figures

Figures reproduced from arXiv: 2607.07091 by the authors.

Figure 1
Figure 1. Overview of the AT-Attn framework. MRI and cognitive-scale features are first encoded and enriched by modality-specific [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Cohort overview and task setup. (A) Cohort construc [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Subgroup gain analysis of the full AT-Attn model rela [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Cohort-level attention analysis of the full AT-Attn model across the five held-out test folds. (A) Mean attention mass by [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 46 canonical work pages

  1. [1]

    2025 Alzheimer’s disease facts and figures,

    Alzheimer’s Association, “2025 Alzheimer’s disease facts and figures,” Alzheimer’s & Dementia, vol. 21, no. 4, p. e70235, Apr. 2025

    work page 2025

  2. [2]

    Mild cognitive impairment: Clinical characterization and outcome,

    R. C. Petersen, G. E. Smith, S. C. Waring, R. J. Ivnik, E. G. Tangalos, and E. Kokmen, “Mild cognitive impairment: Clinical characterization and outcome,”Arch. Neurol., vol. 56, no. 3, pp. 303–308, 1999

    work page 1999

  3. [3]

    The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association work- groups,

    M. S. Albert, S. T. DeKosky, D. Dicksonet al., “The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association work- groups,”Alzheimer’s Dement., vol. 7, no. 3, pp. 270–279, 2011

    work page 2011

  4. [4]

    Alzheimer’s disease neuroimaging initiative (ADNI): clinical characterization,

    R. C. Petersen, P. S. Aisenet al., “Alzheimer’s disease neuroimaging initiative (ADNI): clinical characterization,”Neurology, vol. 74, no. 3, pp. 201–209, 2010

    work page 2010

  5. [5]

    Role of structural MRI in Alzheimer’s disease,

    P. Vemuri and C. R. Jack, Jr., “Role of structural MRI in Alzheimer’s disease,”Alzheimer’s Research & Therapy, vol. 2, no. 4, Art. no. 23, Aug. 2010, doi: 10.1186/alzrt47

    work page doi:10.1186/alzrt47 2010

  6. [6]

    Alzheimer’s disease: Interaction of apolipoprotein E genotype, family history of dementia, gender, education, ethnicity, and age of onset,

    R. Duara, W. W. Barker, R. Lopez-Alberola, D. A. Loewenstein, L. B. Grau, D. Gilchrist, S. Sevush, and P. H. St. George-Hyslop, “Alzheimer’s disease: Interaction of apolipoprotein E genotype, family history of dementia, gender, education, ethnicity, and age of onset,”Neurology, vol. 46, no. 6, pp. 1575–1579, Jun. 1996, doi: 10.1212/WNL.46.6.1575

    work page doi:10.1212/wnl.46.6.1575 1996

  7. [7]

    Patient subtyping via time-aware LSTM networks,

    I. M. Baytas, C. Xiao, X. Zhang, F. Wang, A. K. Jain, and J. Zhou, “Patient subtyping via time-aware LSTM networks,” inProc. ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining (KDD), 2017, pp. 65–74

    work page 2017

  8. [8]

    Recurrent neural networks for multivariate time series with missing values,

    Z. Che, S. Purushotham, K. Cho, D. Sontag, and Y . Liu, “Recurrent neural networks for multivariate time series with missing values,” Scientific Reports, vol. 8, no. 1, p. 6085, 2018

    work page 2018

Show all 46 references
  1. [9]

    Latent ordinary differential equations for irregularly-sampled time series,

    Y . Rubanova, R. T. Q. Chen, and D. K. Duvenaud, “Latent ordinary differential equations for irregularly-sampled time series,” inAdvances in Neural Information Processing Systems 32 (NeurIPS 2019), 2019, pp. 5321–5331

    2019

  2. [10]

    Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups,

    R. A. Sperling, P. S. Aisen, L. A. Beckettet al., “Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups,” Alzheimer’s Dement., vol. 7, no. 3, pp. 280–292, 2011

    2011

  3. [11]

    Hypothetical model of dynamic biomarkers of the Alzheimer’s pathological cascade,

    C. R. Jack Jr., D. S. Knopman, W. J. Jagustet al., “Hypothetical model of dynamic biomarkers of the Alzheimer’s pathological cascade,”Lancet Neurol., vol. 9, no. 1, pp. 119–128, 2010

    2010

  4. [12]

    Tracking patho- physiological processes in Alzheimer’s disease: An updated hypothetical model of dynamic biomarkers,

    C. R. Jack Jr., D. S. Knopman, W. J. Jagustet al., “Tracking patho- physiological processes in Alzheimer’s disease: An updated hypothetical model of dynamic biomarkers,”Lancet Neurol., vol. 12, no. 2, pp. 207– 216, 2013

    2013

  5. [13]

    Preclinical Alzheimer’s disease: Definition, natural history, and diagnostic criteria,

    B. Dubois, H. Hampel, H. H. Feldmanet al., “Preclinical Alzheimer’s disease: Definition, natural history, and diagnostic criteria,”Alzheimer’s Dement., vol. 12, no. 3, pp. 292–323, 2016

    2016

  6. [14]

    A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers,

    C. R. Jack Jr., D. A. Bennett, K. Blennowet al., “A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers,” Neurology, vol. 87, no. 5, pp. 539–547, 2016

    2016

  7. [15]

    Neuroimaging markers for the prediction and early diagnosis of Alzheimer’s disease dementia,

    M. Ewers, R. A. Sperling, W. E. Klunk, M. W. Weiner, and H. Ham- pel, “Neuroimaging markers for the prediction and early diagnosis of Alzheimer’s disease dementia,”Trends Neurosci., vol. 34, no. 8, pp. 430– 442, 2011

    2011

  8. [16]

    MRI and cognitive scores complement each other to ac- curately predict Alzheimer’s dementia 2 to 7 years before clinical onset,

    A. Zandifar, V . S. Fonov, S. Ducharme, S. Belleville, and D. L. Collins, “MRI and cognitive scores complement each other to ac- curately predict Alzheimer’s dementia 2 to 7 years before clinical onset,”NeuroImage: Clinical, vol. 25, Art. no. 102121, 2020, doi: 10.1016/j.nicl....

    2020 doi

  9. [17]

    Multimodal classification of Alzheimer’s disease and mild cognitive impairment,

    D. Zhang, Y . Wang, L. Zhou, H. Yuan, D. Shen, and the ADNI, “Multimodal classification of Alzheimer’s disease and mild cognitive impairment,”NeuroImage, vol. 55, no. 3, pp. 856–867, 2011

    2011

  10. [18]

    Multi-modal multi-task learning for joint pre- diction of multiple regression and classification variables in Alzheimer’s disease,

    D. Zhang and D. Shen, “Multi-modal multi-task learning for joint pre- diction of multiple regression and classification variables in Alzheimer’s disease,”NeuroImage, vol. 59, no. 2, pp. 895–907, 2012

    2012

  11. [19]

    Deep learning-based feature representation for AD/MCI classification,

    H.-I. Suk and D. Shen, “Deep learning-based feature representation for AD/MCI classification,” inProc. Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. (MICCAI), 2013, pp. 583–590

    2013

  12. [20]

    Hierarchical feature representation and multimodal fusion with deep learning for AD/MCI diagnosis,

    H.-I. Suk, S.-W. Lee, and D. Shen, “Hierarchical feature representation and multimodal fusion with deep learning for AD/MCI diagnosis,” NeuroImage, vol. 101, pp. 569–582, 2014

    2014

  13. [21]

    Multimodal neuroimaging feature learning for multiclass diagnosis of Alzheimer’s disease,

    S. Liu, S. Liu, W. Cai, H. Che, S. Pujol, R. Kikinis, D. Feng, and M. J. Fulham, “Multimodal neuroimaging feature learning for multiclass diagnosis of Alzheimer’s disease,”IEEE Trans. Biomed. Eng., vol. 62, no. 4, pp. 1132–1140, 2015, doi: 10.1109/TBME.2014.2372011

    2015 doi

  14. [22]

    Canonical feature selection for joint regression and multi-class identification in Alzheimer’s disease diagnosis,

    X. Zhu, H.-I. Suk, and D. Shen, “Canonical feature selection for joint regression and multi-class identification in Alzheimer’s disease diagnosis,”Brain Imaging Behav., vol. 10, no. 3, pp. 818–828, 2016

    2016

  15. [23]

    Label-aligned multi-task feature learning for multimodal classification of Alzheimer’s disease and mild cognitive impairment,

    C. Zu, B. Jie, M. Liu, S. Chen, D. Shen, and D. Zhang, “Label-aligned multi-task feature learning for multimodal classification of Alzheimer’s disease and mild cognitive impairment,”Brain Imaging Behav., vol. 10, no. 4, pp. 1148–1159, 2016

    2016

  16. [24]

    Identification of Alzheimer’s disease and mild cognitive impairment using multimodal sparse hierarchical extreme learning machine,

    J. Kim and B. Lee, “Identification of Alzheimer’s disease and mild cognitive impairment using multimodal sparse hierarchical extreme learning machine,”Hum. Brain Mapp., vol. 39, no. 9, pp. 3728–3741, 2018

    2018

  17. [25]

    Automated classification of Alzheimer’s disease and mild cognitive impairment using a single MRI and deep neural networks,

    S. Basaia, F. Agosta, L. Wagneret al., “Automated classification of Alzheimer’s disease and mild cognitive impairment using a single MRI and deep neural networks,”NeuroImage Clin., vol. 21, Art. no. 101645, 2019

    2019

  18. [26]

    Development and validation of an interpretable deep learning framework for Alzheimer’s disease classification,

    S. Qiu, P. S. Joshi, M. I. Milleret al., “Development and validation of an interpretable deep learning framework for Alzheimer’s disease classification,”Brain, vol. 143, no. 6, pp. 1920–1933, 2020

    1920

  19. [27]

    A multilayer multimodal detection and prediction model based on explainable artificial intelligence for Alzheimer’s disease,

    S. El-Sappagh, J. M. Alonso, S. M. R. Islam, A. M. Sultan, and K. S. Kwak, “A multilayer multimodal detection and prediction model based on explainable artificial intelligence for Alzheimer’s disease,”Sci. Rep., vol. 11, Art. no. 2660, 2021, doi: 10.1038/s41598-021-82098-3

    2021 doi

  20. [28]

    DiaMond: De- mentia diagnosis with multi-modal vision Transformers using MRI and PET,

    Y . Li, M. Ghahremani, Y . Wally, and C. Wachinger, “DiaMond: De- mentia diagnosis with multi-modal vision Transformers using MRI and PET,” inProc. WACV, 2025, pp. 107–116

    2025

  21. [29]

    Multistage alignment and fusion for multimodal multiclass Alzheimer’s disease diagnosis,

    S. Huang, L. Zhong, and Y . Shi, “Multistage alignment and fusion for multimodal multiclass Alzheimer’s disease diagnosis,” inMedical Image Computing and Computer Assisted Intervention – MICCAI 2025, vol. 15974. Springer Nature Switzerland, 2025, pp. 375–385

    2025

  22. [30]

    Alzheimer’s disease recognition based on adaptive graph normalization flow for incomplete multimodal data fusion,

    Y . Li, Y . Dong, Y . Wu, H. Yan, and L. Gao, “Alzheimer’s disease recognition based on adaptive graph normalization flow for incomplete multimodal data fusion,” inMedical Image Computing and Computer Assisted Intervention – MICCAI 2025, vol. 15967. Springer Nature Switzerland...

    2025

  23. [31]

    UniCross: Balanced multimodal learning for Alzheimer’s disease diagnosis by uni-modal separation and metadata-guided cross-modal interaction,

    L. Yin, C. Ye, T. Liu, J. Wu, and T. Yan, “UniCross: Balanced multimodal learning for Alzheimer’s disease diagnosis by uni-modal separation and metadata-guided cross-modal interaction,” inMedical Image Computing and Computer Assisted Intervention – MICCAI 2025, vol. 15974. Spr...

    2025

  24. [32]

    A review on neuroimaging-based classification studies and associated feature extraction methods for Alzheimer’s disease and its prodromal stages,

    S. Rathore, M. Habes, M. A. Iftikhar, A. Shacklett, and C. Davatzikos, “A review on neuroimaging-based classification studies and associated feature extraction methods for Alzheimer’s disease and its prodromal stages,”NeuroImage, vol. 155, pp. 530–548, 2017

    2017

  25. [33]

    Machine learning of neuroimaging for assisted diagnosis of cognitive impairment and dementia: A systematic review,

    E. Pellegrini, L. Ballerini, M. C. V . Hernandezet al., “Machine learning of neuroimaging for assisted diagnosis of cognitive impairment and dementia: A systematic review,”Alzheimer’s Dement. Diagn. Assess. Dis. Monit., vol. 10, pp. 519–535, 2018

    2018

  26. [34]

    A distributed multitask multimodal approach for the prediction of Alzheimer’s disease in a longitudinal study,

    S. Tabarestani, M. Aghili, M. Eslamiet al., “A distributed multitask multimodal approach for the prediction of Alzheimer’s disease in a longitudinal study,”NeuroImage, vol. 206, Art. no. 116317, 2020

    2020

  27. [35]

    Predicting Alzheimer’s disease progression using deep recurrent neural networks,

    M. Nguyen, T. He, L. An, D. C. Alexander, J. Feng, and B. T. T. Yeo, “Predicting Alzheimer’s disease progression using deep recurrent neural networks,”NeuroImage, vol. 222, Art. no. 117203, 2020

    2020

  28. [36]

    Phased LSTM: Accelerating recur- rent network training for long or event-based sequences,

    D. Neil, M. Pfeiffer, and S.-C. Liu, “Phased LSTM: Accelerating recur- rent network training for long or event-based sequences,” inAdvances in Neural Information Processing Systems 29 (NeurIPS 2016), 2016, pp. 3882–3890

    2016

  29. [37]

    TA-RNN: an attention-based time-aware recurrent neural network architecture for electronic health records,

    M. Al Olaimat and S. Bozdag, “TA-RNN: an attention-based time-aware recurrent neural network architecture for electronic health records,” Bioinformatics, vol. 40, no. Suppl. 1, pp. i169–i179, 2024

    2024

  30. [38]

    BRITS: Bidirectional recurrent imputation for time series,

    W. Cao, D. Wang, J. Li, H. Zhou, L. Li, and Y . Li, “BRITS: Bidirectional recurrent imputation for time series,” inAdvances in Neural Information Processing Systems 31 (NeurIPS 2018), 2018, pp. 6775–6785

    2018

  31. [39]

    GRU-ODE-Bayes: Continuous modeling of sporadically-observed time series,

    E. De Brouwer, J. Simm, A. Arany, and Y . Moreau, “GRU-ODE-Bayes: Continuous modeling of sporadically-observed time series,” inAdvances in Neural Information Processing Systems 32 (NeurIPS 2019), 2019, pp. 7379–7390

    2019

  32. [40]

    Neural controlled differential equations for irregular time series,

    P. Kidger, J. Morrill, J. Foster, and T. Lyons, “Neural controlled differential equations for irregular time series,” inAdvances in Neural Information Processing Systems 33 (NeurIPS 2020), 2020, pp. 6696– 6707

    2020

  33. [41]

    Multi-time attention networks for irregularly sampled time series,

    S. N. Shukla and B. M. Marlin, “Multi-time attention networks for irregularly sampled time series,” inProc. Int. Conf. Learn. Represent. (ICLR), 2021

    2021

  34. [42]

    Multimodal machine learning: A survey and taxonomy,

    T. Baltrusaitis, C. Ahuja, and L.-P. Morency, “Multimodal machine learning: A survey and taxonomy,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 41, no. 2, pp. 423–443, 2019

    2019

  35. [43]

    A survey on deep multimodal learning for computer vision: Advances, trends, applications, and datasets,

    K. Bayoudh, R. Knani, F. Hamdaoui, and A. Mtibaa, “A survey on deep multimodal learning for computer vision: Advances, trends, applications, and datasets,”Vis. Comput., vol. 38, pp. 2939–2970, 2022

    2022

  36. [44]

    What makes training multi- modal classification networks hard?

    W. Wang, D. Tran, and M. Feiszli, “What makes training multi- modal classification networks hard?” inProc. IEEE/CVF CVPR, 2020, pp. 12695–12705

    2020

  37. [45]

    Balanced multimodal learning via on-the-fly gradient modulation,

    X. Peng, Y . Wei, A. Deng, D. Wang, and D. Hu, “Balanced multimodal learning via on-the-fly gradient modulation,” inProc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), 2022, pp. 8238–8247

    2022

  38. [46]

    Med3D: Transfer learning for 3D medical image analysis,

    S. Chen, K. Ma, and Y . Zheng, “Med3D: Transfer learning for 3D medical image analysis,”arXiv preprint arXiv:1904.00625, 2019

    1904 arXiv

Pith tools

Reviewed July 9, 2026 · model on record in the stance chip above.