Pith. sign in

REVIEW 3 major objections 4 minor 48 references

Dementia Etiology Diagnosis via Collaborative Meta Knowledge Enhancement

T0 review · 3 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read A single diagnostic model can handle the heterogeneity of seven dementia cohorts by explicitly encoding scan protocol, cohort source, and modality availability as learned tokens, regularized by a trust-region reference model.

desk verdict A serious multi-center dementia diagnosis paper with strong empirical results, but the trust-region anchor is self-referential and the OOD protocol has an unspecified gap; send to review. read the letter →

arxiv 2607.22770 v1 pith:O4OQ4AXK submitted 2026-07-24 cs.LG cs.CV

Siyuan Du , Mengxi Chen , Xinyang Jiang , Zilong Wang , Jiangchao Yao , Dongsheng Li , Ya Zhang , Lili Qiu
show 1 more author
Yanfeng Wang
This is my paper
classification cs.LGcs.CV
keywords dementiaetiologydiagnosismulti-centerlearningdataheterogeneitymeta-knowledgeinjectiontrust-regionoptimizationspuriouscorrelationmultimodalfusionTransformer
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 show that a single diagnostic model can handle the messy reality of multi-center dementia data—where scanners, scan protocols, population mixes, and even which modalities are available differ from site to site—if the model is told, explicitly, what kind of site and scan it is looking at. The proposed framework, COME, encodes acquisition-protocol descriptions, cohort identifiers, and modality-presence indicators as learned tokens and injects them into a Transformer, then uses a reference model trained without those tokens to keep the enhanced model from latching onto site-specific shortcuts. On seven independent cohorts the method reports a mean macro-averaged AUC of 85.62%, a 4.29-point gain over the strongest baseline, and better cross-center and cross-sequence generalization. The authors argue this makes real-world, explainable dementia diagnostics more feasible.

What carries the argument

Meta-knowledge enhancement: three types of heterogeneity-aware embeddings—acquisition-aware (LLM-standardized sequence descriptions encoded by a medical text model), cohort-aware (trainable per-source tokens), and modality-coverage (present/missing tokens per modality)—are injected into a shared Transformer via cross-attention to refine image features before multimodal fusion. The trust-region constraint is the counterweight: a reference model without these embeddings, refreshed every k epochs, anchors dual-level alignment—a minibatch contrastive loss on pooled patient representations (global) and a cosine-affinity MSE on patch self-similarity matrices (local)—preventing the meta-injection f

What would settle it

Keep the reference model frozen at the end of the initialization phase (no periodic θref ← θmain refreshes) and compare cross-center AUC to the default schedule; if performance does not drop, the periodic refresh is not what prevents spurious correlations, and if the reference and main representations become near-identical, the constraint is vacuous.

Watch

Extended reading notes

Core claim

COME claims that the obstacle to scale-up multi-center dementia diagnosis is not simply lack of data but unmodeled heterogeneity: acquisition protocols, cohort demographics, and modality coverage encode non-biological variation that pooled models tend to memorize. The paper's solution is to make that heterogeneity explicit. Acquisition sequence names are standardized by a large language model into technical descriptions, then encoded by a medical text encoder; cohort source and modality presence/absence become trainable tokens. These meta-knowledge embeddings are injected through cross-attention refinement layers so image features are modulated conditioned on the source and protocol. To prev

Load-bearing premise

The trust-region reference model is assumed to represent only core clinical signals, but because it is periodically overwritten with the main model's parameters every k epochs, after the first update it is a lagged copy of the meta-enhanced model rather than an independent meta-free anchor; if that anchor already contains site- and modality-specific biases, the constraint cannot remove them.

Editorial extensions

If this is right

  • Pooling multi-center dementia data with explicit protocol/source/modality tokens yields better in-domain diagnosis than single-center or multi-task baselines, especially for rare etiologies and small cohorts.
  • The gain transfers to unseen centers and unseen MRI sequences, so models trained this way are more likely to work when deployed at a new hospital.
  • Constraining meta-injection with the reference model is necessary: unconstrained injection can hurt, so the regularization, not just the extra information, drives robustness.
  • Model predictions track established biomarkers and clinical severity (CDR, amyloid/tau), suggesting the learned representations are biologically meaningful, not just statistically tuned.
  • The two-phase training schedule (diagnosis-first anchor, then constrained enhancement) is important for the benefit; removing or altering it degrades performance.

Reading between the lines

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

  • If the trust-region anchor is periodically overwritten with the enhanced model's weights, the reference may drift toward the same spurious correlations the constraint is meant to remove; the paper does not compare against a never-refreshed reference, so the mechanism's causal contribution is not fully isolated.
  • The method's reliance on an LLM to standardize sequence names adds an offline step whose errors (if any) would propagate; the paper's radiologist check is a one-time audit, not a scalable guarantee for new sites.
  • The approach suggests a general recipe for heterogeneity in medical imaging beyond dementia: any multi-center pooled training could benefit from explicit acquisition and cohort tokens plus an unenhanced anchor, provided the anchor stays clean.
  • A natural testable extension: vary the refresh interval k and reference initialization to see at what point the constraint becomes vacuous (reference ≈ main) and OOD performance degrades.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The manuscript proposes COME, a Transformer-based framework for multi-center dementia etiology diagnosis. It injects three types of meta-knowledge embeddings—acquisition-sequence, cohort-source, and modality-coverage—into the image and clinical feature streams, and regularizes training with a trust-region objective built on a reference model. The authors evaluate on seven cohorts for in-domain classification and on leave-one-center-out and cross-sequence settings for out-of-domain generalization, reporting state-of-the-art macro-averaged AUC with a 4.29-point gain over the strongest baseline. They also provide biomarker- and CDR-based validation of model predictions. The central mechanistic claim is that the trust-region anchor prevents spurious correlations by keeping the meta-enhanced model close to a representation space learned exclusively from core clinical inputs and free of heterogeneity-induced biases.

Significance. If the mechanism were supported, this would be a practically valuable contribution to multi-center medical imaging: it tackles a real deployment problem (site and modality heterogeneity) with an explicit modeling strategy and demonstrates results across seven public cohorts, including OOD evaluation. The paper's strengths include the breadth of the multi-center evaluation, the inclusion of cross-sequence generalization, and the external validation against amyloid, tau, FDG, and DaTscan biomarkers and CDR scores. The ablations and hyperparameter sensitivity analyses are useful. However, the load-bearing explanation for why the method works—the trust-region anchor being free of heterogeneity-induced biases—is not supported by the training procedure as written, and the leave-one-center-out protocol leaves a key inference detail unspecified.

major comments (3)
  1. [Sec. III-C and III-D, Eq. (11)] The trust-region anchor is not independently trained. The reference model is initialized from the no-meta phase, but then θref ← θmain every k=2 epochs throughout the constrained phase. After the first refresh, θref is a lagged copy of the meta-enhanced main model optimized with L = Lcls + λ(Lg+Ll). Thus the reference model's weights have been shaped by meta-knowledge gradients, even though it receives no meta tokens at inference. The claim in Sec. III-C that the trust region is 'the space of representations learned exclusively from core clinical inputs... free from heterogeneity-induced biases' is therefore not established. The reported gains may arise from temporal averaging or other effects, but the spurious-correlation-prevention mechanism is unsupported. Please retrain an independent reference model with Lcls only (never overwritten by the main model) and compare, or revise the mech
  2. [Sec. IV-F1 and Sec. III-B1b] The cross-center evaluation is leave-one-center-out, but cohort-aware embeddings P^s_cohort are trainable per source. For a held-out center, no such embedding was trained. The paper does not state how the model obtains a cohort embedding for the held-out center at inference—whether it is omitted, averaged over training sources, or set to a learned default. This choice changes the input distribution and directly affects the validity of the reported OOD results. Specify the exact protocol and, if an embedding is synthesized, justify it.
  3. [Table VI and Sec. IV-G1] The 'No Enhancement' ablation is described as meta information not used during inference. But the model in that ablation was trained with meta-knowledge tokens and trust-region constraints; removing the tokens only at inference changes the input distribution and may not reflect what a model trained without meta-knowledge would learn. To support the claim that meta-knowledge injection itself improves robustness, report a variant trained without meta-knowledge embeddings from the start, alongside the inference-only removal variant.
minor comments (4)
  1. [Fig. 6 caption] The caption contains garbled text: 'AℬTa u' should be 'Aβ and tau'. Please fix.
  2. [Author affiliations] 'Y a Zhang' has an extra space; should be 'Ya Zhang'.
  3. [Sec. III-B2, Eq. (5)] N is introduced as the patch token number after fusion, but the notation is overloaded: earlier N_p was used for image patches. Please clarify the dimensions of the fused sequence.
  4. [Sec. IV-E] The statistical test is a paired two-sided t-test over five seeds. Given only five seeds, a paired permutation test or reporting effect sizes with confidence intervals would strengthen the claim.

Circularity Check

1 steps flagged · score 4.0 of 10

Trust-region reference model is periodically copied from the meta-enhanced main model, so the claimed bias-free anchor is self-referential; the mechanism for spurious-correlation prevention is unsupported, though the empirical results remain independent.

  1. self definitional [Section III-C (Trust-region Constrained Optimization) and Section III-D (Training Process), including Eq. (11) and the θref ← θmain hard-copy schedule]
    "The trust-region is defined as the space of representations learned exclusively from core clinical inputs (imaging and clinical data), free from heterogeneity-induced biases. ... we instantiate the trust-region optimization via a reference model (identical architecture to the main model but without meta knowledge enhancement) that serves as an intermittently refreshed trust-region anchor through hard parameter updates every k training epochs: θref ← θmain when t mod k = 0 ... its parameters are refreshed by θref ← θmain every k epochs, with k=2 in the main experiments."

    The constraint's anchor is defined as a representation learned exclusively from core clinical inputs and therefore free of heterogeneity-induced biases. But the implementation defines the anchor as a lagged copy of the main model: after the initialization phase, θref is periodically set equal to θmain, where θmain is optimized with L = Lcls + λ(Lg + Ll) while receiving meta-knowledge tokens. Hence, on every refresh, the reference inherits whatever site/modality/missingness correlations the meta-enhanced main model has already learned, and Lg/Ll pull the main model toward that contaminated copy. The trust-region therefore reduces to agreement with the model's own past, not to an independent clinical-only no-bias representation; the claimed spurious-correlation-prevention mechanism does not

full rationale

There is no equation-level circularity in the conventional sense: COME's in-domain and OOD numbers are measured against external baselines on held-out centers/sequences, and the biomarker/CDR correlations are independent external validations that were not used to fit constants. The self-citations ([8], [26], [29]) appear only in related-work motivation and are not load-bearing. The one substantive circularity is the trust-region mechanism: the reference model is asserted to define a bias-free clinical-only trust region, but the implementation refreshes θref from the meta-enhanced main model every k=2 epochs. After the first refresh, the anchor is not independent of the very meta-knowledge-enhanced training it is supposed to constrain; the global and local alignment losses then enforce self-agreement with a lagged version of the model rather than with an unbiased representation. This undermines the paper's mechanistic explanation for OOD robustness, though not the empirical claims themselves. Score 4 reflects that the central empirical content is still externally grounded, while one load-bearing mechanistic argument is partly circular as written.

Assumptions & free parameters 6 free parameters · 5 assumptions · 4 invented entities

The central claims rest on hand-selected hyperparameters, trainable meta-knowledge tokens, and domain assumptions about label accuracy, LLM standardization, and the trust-region reference model. The most consequential is the reference-model assumption, because the mechanism of spurious-correlation prevention is only as strong as the anchor, and the anchor is a lagged copy of the model it constrains.

free parameters (6)
  • trust-region loss weight λ = 0.2
    Selected by ablation; λ=0.1/0.3 gave slightly lower performance and λ=1 substantially reduced AUC-m (Sec. IV-G.2).
  • reference update interval k = 2 epochs
    Chosen as default; k=1 and k=3 performed comparably (Sec. IV-G.2), indicating hand selection.
  • meta-knowledge refinement depth J = 2
    Ablated: J=1 worse, J=3/4 inconsistent, J=2 chosen (Sec. IV-G.4).
  • contrastive temperature τ
    Used in the global alignment loss Eq. 7, but its value is never reported; affects constraint strength.
  • per-source cohort embeddings = 7 trainable vectors (one per cohort)
    Trainable tokens P^s_cohort (Sec. III-B.1.b) absorb source-specific label/artifact distribution; they may encode spurious site signals.
  • modality-coverage embeddings = 4 trainable vectors (present/missing for MRI and PET)
    Trainable tokens P^m_mod and bar-P^m_mod explicitly encode missingness, which can correlate with diagnosis and act as a shortcut.
assumptions (5)
  • domain assumption Source-documented clinical diagnosis fields are accurate enough to serve as ground-truth dementia etiologies after harmonization
    Label harmonization in Appendix II relies on NACC UDS fields, ADNI cohort assignments, etc.; errors in clinical diagnosis propagate to the target labels.
  • domain assumption DeepSeek V3 standardization plus radiologist review yields correct technical descriptions for all 247 sequence names
    Acquisition-aware embeddings are derived from LLM-generated descriptions (Sec. III-B.1.a, Appendix I); incorrect descriptions mislead the model.
  • ad hoc to paper A reference model trained on the same multi-center data without meta tokens is free of heterogeneity-induced spurious correlations
    Sec. III-C defines the trust region as representations 'free from heterogeneity-induced biases', but no evidence shows the no-meta model lacks such biases, and it is periodically copied from the main model (Sec. III-D).
  • domain assumption Zero-imputation with observation masks for numerical clinical variables is a valid missingness strategy
    Clinical variable coverage ranges from about 10% to 75% per cohort (Table III); the fusion and trust-region losses assume masked zeros do not distort diagnosis.
  • ad hoc to paper Leave-one-center-out evaluation can be run for a held-out center without a trained source embedding
    No description is given of how P^s_cohort is set for an unseen site in Sec. IV-F.1; the OOD claim depends on this missing detail.
invented entities (4)
  • Acquisition-aware sequence embeddings (P^MRI_seq, P^PET_seq)
    purpose: Condition image tokens on standardized scan-sequence semantics
    Internal representation; Fig. 4(a) shows semantic grouping but no external handle beyond the paper.
  • Cohort-aware source embeddings (P^s_cohort)
    purpose: Capture site-specific distribution biases
    Learned per source; may encode site-label shortcuts; no external validation.
  • Modality-coverage embeddings (P^m_mod, bar-P^m_mod)
    purpose: Inform the model about missing modalities
    Trainable tokens; missingness patterns can correlate with diagnosis and act as shortcuts.
  • Reference model used as trust-region anchor
    purpose: Constrain meta-enhanced representations to a clinically valid region
    Not an independent anchor: it is periodically overwritten with main-model parameters (Sec. III-D).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Dementia Etiology Diagnosis via Collaborative Meta Knowledge Enhancement." pith.science (2026). https://pith.science/paper/O4OQ4AXK

@misc{pith2026260722770,
  author       = {Pith},
  title        = {Pith review of: Dementia Etiology Diagnosis via Collaborative Meta Knowledge Enhancement},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O4OQ4AXK}},
  note         = {Machine review of arXiv:2607.22770}
}
read the original abstract

Although artificial intelligence (AI) has shown promising performance in several medical tasks, accurate dementia etiology diagnosis with AI remains challenging due to complex overlapping symptoms among diseases. Scaling up the dataset size by combining the cross-center samples may bring a gain in the pursuit of performance, while the inherent data heterogeneity across centers or populations induces the conflict. Conventional multi-task learning paradigms offer a promising framework; however, they fail to consider critical meta information (e.g., site-specific acquisition and modality availability) to combat the heterogeneity. To address this challenge, we propose a Collaborative Meta Knowledge Enhancement (COME) framework for dementia etiology diagnosis, which injects multi-center acquisition semantics, source identifiers, and modality indicators as heterogeneity-aware embeddings into a unified Transformer architecture for scale-up training, enabling explicit modeling of heterogeneity. Besides, a trust-region constrained optimization scheme is designed to regularize the model from spurious correlations during training through a reference model. Across seven independent cohorts, our method achieves state-of-the-art in-domain performance with a mean macro-averaged AUC of 85.62% and a 4.29-point gain over the strongest baseline, while maintaining superior out-of-domain generalization under both cross-center and cross-sequence evaluations. Extensive validation also confirms the alignment between model predictions and established biomarkers (amyloid, tau) and clinical severity, highlighting the potential of COME to enable robust and interpretable dementia diagnostics in real-world settings.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 8 linked inside Pith

  1. [1]

    Global status report on the public health response to dementia,

    W. H. Organization et al., “Global status report on the public health response to dementia,” 2021

    2021

  2. [2]

    Accuracy of the clinical diagnosis of alzheimer disease at national institute on aging alzheimer disease centers, 2005–2010,

    T. G. Beach, S. E. Monsell, L. E. Phillips, and W. Kukull, “Accuracy of the clinical diagnosis of alzheimer disease at national institute on aging alzheimer disease centers, 2005–2010,” Journal of neuropathology and experimental neurology, vol. 71, no. 4, pp. 266–273, 2012

    2005

  3. [3]

    Amyloid-related imaging abnormalities (aria): radiological, biological and clinical characteristics,

    H. Hampel, A. Elhage, M. Cho, L. G. Apostolova, J. A. Nicoll, and A. Atri, “Amyloid-related imaging abnormalities (aria): radiological, biological and clinical characteristics,” Brain, vol. 146, no. 11, pp. 4414– 4424, 2023. AUTHORet al.: PREPARATION OF PAPERS FOR IEEE TRANSACTIONS ON MEDICAL IMAGING 13

    2023

  4. [4]

    Deep convolutional neural network based classification of alzheimer’s disease using mri data,

    A. Nawaz, S. M. Anwar, R. Liaqat, J. Iqbal, U. Bagci, and M. Majid, “Deep convolutional neural network based classification of alzheimer’s disease using mri data,” in 2020 IEEE 23rd International Multitopic Conference (INMIC). IEEE, 2020, pp. 1–6

    2020

  5. [5]

    Multimodal deep learning for alzheimer’s disease dementia assessment,

    S. Qiu, M. I. Miller, P. S. Joshi, J. C. Lee, C. Xue, Y . Ni, Y . Wang, I. De Anda-Duran, P. H. Hwang, J. A. Cramer et al., “Multimodal deep learning for alzheimer’s disease dementia assessment,” Nature communications, vol. 13, no. 1, p. 3404, 2022

    2022

  6. [6]

    Ai-based differential diagnosis of dementia etiologies on multimodal data,

    C. Xue, S. S. Kowshik, D. Lteif, S. Puducheri, V . H. Jasodanand, O. T. Zhou, A. S. Walia, O. B. Guney, J. D. Zhang, S. Po ´esy et al., “Ai-based differential diagnosis of dementia etiologies on multimodal data,” Nature Medicine, vol. 30, no. 10, pp. 2977–2989, 2024

    2024

  7. [7]

    Discovery of novel csf biomarkers to predict progression in dementia using machine learning,

    D. Gogishvili, E. M. Vromen, S. Koppes-den Hertog, A. W. Lemstra, Y . A. Pijnenburg, P. J. Visser, B. M. Tijms, M. Del Campo, S. Abeln, C. E. Teunissen et al., “Discovery of novel csf biomarkers to predict progression in dementia using machine learning,” Scientific Reports, vol. 13, no. 1, p. 6531, 2023

    2023

  8. [8]

    Lorkd: Low-rank knowledge decomposition for medical foundation models,

    H. Li, Y . Zhou, Z. Zhao, S. Du, J. Yao, W. Xie, Y . Zhang, and Y . Wang, “Lorkd: Low-rank knowledge decomposition for medical foundation models,” arXiv preprint arXiv:2409.19540, 2024

    arXiv 2024

Show all 48 references
  1. [9]

    Scaling up visual and vision-language representation learning with noisy text supervision,

    C. Jia, Y . Yang, Y . Xia, Y .-T. Chen, Z. Parekh, H. Pham, Q. Le, Y .-H. Sung, Z. Li, and T. Duerig, “Scaling up visual and vision-language representation learning with noisy text supervision,” in International conference on machine learning. PMLR, 2021, pp. 4904–4916

    2021

  2. [10]

    Multitask learning: A knowledge-based source of inductive bias1,

    R. Caruana, “Multitask learning: A knowledge-based source of inductive bias1,” in Proceedings of the Tenth International Conference on Machine Learning, 1993, pp. 41–48

    1993

  3. [11]

    Low resource dependency parsing: Cross-lingual parameter sharing in a neural network parser,

    L. Duong, T. Cohn, S. Bird, and P. Cook, “Low resource dependency parsing: Cross-lingual parameter sharing in a neural network parser,” in Proceedings of the 53rd annual meeting of the Association for Computational Linguistics and the 7th international joint conference on natu...

    2015

  4. [12]

    Outrageously large neural networks: The sparsely-gated mixture-of-experts layer,

    N. Shazeer, A. Mirhoseini, K. Maziarz, A. Davis, Q. Le, G. Hinton, and J. Dean, “Outrageously large neural networks: The sparsely-gated mixture-of-experts layer,” arXiv preprint arXiv:1701.06538, 2017

    2017 arXiv

  5. [13]

    Learning multiple tasks with multilinear relationship networks,

    M. Long, Z. Cao, J. Wang, and P. S. Yu, “Learning multiple tasks with multilinear relationship networks,” Advances in neural information processing systems, vol. 30, 2017

    2017

  6. [14]

    Modeling task relationships in multi-task learning with multi-gate mixture-of-experts,

    J. Ma, Z. Zhao, X. Yi, J. Chen, L. Hong, and E. H. Chi, “Modeling task relationships in multi-task learning with multi-gate mixture-of-experts,” in Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining, 2018, pp. 1930–1939

    2018

  7. [15]

    Independent component alignment for multi-task learning,

    D. Senushkin, N. Patakin, A. Kuznetsov, and A. Konushin, “Independent component alignment for multi-task learning,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023, pp. 20 083–20 093

    2023

  8. [16]

    Disentangling heterogeneity in alzheimer’s disease and related dementias using data-driven methods,

    M. Habes, M. J. Grothe, B. Tunc, C. McMillan, D. A. Wolk, and C. Davatzikos, “Disentangling heterogeneity in alzheimer’s disease and related dementias using data-driven methods,” Biological psychiatry, vol. 88, no. 1, pp. 70–82, 2020

    2020

  9. [17]

    Segmentation of brain tumors using deeplabv3+,

    A. Roy Choudhury, R. Vanguri, S. R. Jambawalikar, and P. Kumar, “Segmentation of brain tumors using deeplabv3+,” inBrainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries: 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Gra...

    2018

  10. [18]

    Springer, 2019, pp. 154–167

    2019

  11. [19]

    Mri-based multi-task decoupling learning for alzheimer’s disease detection and mmse score prediction: A multi-site validation,

    X. Tian, J. Liu, H. Kuang, Y . Sheng, J. Wang, A. D. N. Initiative et al., “Mri-based multi-task decoupling learning for alzheimer’s disease detection and mmse score prediction: A multi-site validation,” arXiv preprint arXiv:2204.01708, 2022

    2022 arXiv

  12. [20]

    Multi-task learning for alzheimer’s disease diagnosis and mini-mental state examination score prediction,

    J. Liu, X. Tian, H. Lin, H.-D. Li, and Y . Pan, “Multi-task learning for alzheimer’s disease diagnosis and mini-mental state examination score prediction,” Big Data Mining and Analytics, vol. 7, no. 3, pp. 828–842, 2024

    2024

  13. [21]

    A lifelong learning approach to brain mr segmentation across scanners and protocols,

    N. Karani, K. Chaitanya, C. Baumgartner, and E. Konukoglu, “A lifelong learning approach to brain mr segmentation across scanners and protocols,” in Med. Image Comput. Comput. Assist. Interv. Springer, 2018, pp. 476–484

    2018

  14. [22]

    A cross- stitch architecture for joint registration and segmentation in adaptive radiotherapy,

    L. Beljaards, M. S. Elmahdy, F. Verbeek, and M. Staring, “A cross- stitch architecture for joint registration and segmentation in adaptive radiotherapy,” in Medical Imaging with Deep Learning. PMLR, 2020, pp. 62–74

    2020

  15. [23]

    A foundation model for brain lesion segmentation with mixture of modality experts,

    X. Zhang, N. Ou, B. D. Basaran, M. Visentin, M. Qiao, R. Gu, C. Ouyang, Y . Liu, P. M. Matthews, C. Ye et al., “A foundation model for brain lesion segmentation with mixture of modality experts,” in Med. Image Comput. Comput. Assist. Interv. Springer, 2024, pp. 379–389

    2024

  16. [24]

    Multimodal gated mixture of experts using whole slide image and flow cytometry for multiple instance learning classification of lymphoma,

    N. Hashimoto, H. Hanada, H. Miyoshi, M. Nagaishi, K. Sato, H. Hontani, K. Ohshima, and I. Takeuchi, “Multimodal gated mixture of experts using whole slide image and flow cytometry for multiple instance learning classification of lymphoma,” Journal of Pathology Informatics, vol...

    2024

  17. [25]

    Task adaptive parameter sharing for multi- task learning,

    M. Wallingford, H. Li, A. Achille, A. Ravichandran, C. Fowlkes, R. Bhotika, and S. Soatto, “Task adaptive parameter sharing for multi- task learning,” in IEEE Conf. Comput. Vis. Pattern Recog., 2022, pp. 7561–7570

    2022

  18. [26]

    Dselect-k: Differentiable selection in the mixture of experts with applications to multi-task learning,

    H. Hazimeh, Z. Zhao, A. Chowdhery, M. Sathiamoorthy, Y . Chen, R. Mazumder, L. Hong, and E. Chi, “Dselect-k: Differentiable selection in the mixture of experts with applications to multi-task learning,” Advances in Neural Information Processing Systems, vol. 34, pp. 29 335–29 ...

    2021

  19. [27]

    Reprogramming distillation for medical foundation models,

    Y . Zhou, S. Du, H. Li, J. Yao, Y . Zhang, and Y . Wang, “Reprogramming distillation for medical foundation models,” in Med. Image Comput. Comput. Assist. Interv. Springer, 2024, pp. 533–543

    2024

  20. [28]

    Reducing reliance on spurious features in medical image classification with spatial specificity,

    K. Saab, S. Hooper, M. Chen, M. Zhang, D. Rubin, and C. R ´e, “Reducing reliance on spurious features in medical image classification with spatial specificity,” in Machine Learning for Healthcare Conference. PMLR, 2022, pp. 760–784

    2022

  21. [29]

    Smart: Towards pre-trained missing-aware model for patient health status prediction,

    Z. Yu, C. Xu, Y . Jin, Y . Wang, and J. Zhao, “Smart: Towards pre-trained missing-aware model for patient health status prediction,” Advances in Neural Information Processing Systems, vol. 37, pp. 63 986–64 009, 2024

    2024

  22. [30]

    Exploring training on heterogeneous data with mixture of low-rank adapters,

    Y . Zhou, Z. Zhao, H. Li, S. Du, J. Yao, Y . Zhang, and Y . Wang, “Exploring training on heterogeneous data with mixture of low-rank adapters,” arXiv preprint arXiv:2406.09679, 2024

    2024 arXiv

  23. [31]

    Deepseek-v3 technical report,

    A. Liu, B. Feng, B. Xue, B. Wang, B. Wu, C. Lu, C. Zhao, C. Deng, C. Zhang, C. Ruan et al., “Deepseek-v3 technical report,” arXiv preprint arXiv:2412.19437, 2024

    2024 arXiv

  24. [32]

    A generalist medical language model for disease diagnosis assistance,

    X. Liu, H. Liu, G. Yang, Z. Jiang, S. Cui, Z. Zhang, H. Wang, L. Tao, Y . Sun, Z. Song et al., “A generalist medical language model for disease diagnosis assistance,” Nature medicine, vol. 31, no. 3, pp. 932–942, 2025

    2025

  25. [33]

    Prefix conditioning unifies language and label supervision,

    K. Saito, K. Sohn, X. Zhang, C.-L. Li, C.-Y . Lee, K. Saenko, and T. Pfister, “Prefix conditioning unifies language and label supervision,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023, pp. 2861–2870

    2023

  26. [34]

    Nia-aa research framework: toward a biological definition of alzheimer’s disease,

    C. R. Jack Jr, D. A. Bennett, K. Blennow, M. C. Carrillo, B. Dunn, S. B. Haeberlein, D. M. Holtzman, W. Jagust, F. Jessen, J. Karlawish et al., “Nia-aa research framework: toward a biological definition of alzheimer’s disease,” Alzheimer’s & dementia, vol. 14, no. 4, pp. 535–562, 2018

    2018

  27. [35]

    Representation learning with contrastive predictive coding,

    A. v. d. Oord, Y . Li, and O. Vinyals, “Representation learning with contrastive predictive coding,” arXiv preprint arXiv:1807.03748, 2018

    2018 arXiv

  28. [36]

    The national alzheimer’s coordinating center (nacc) database: an alzheimer disease database,

    D. L. Beekly, E. M. Ramos, G. van Belle, W. Deitrich, A. D. Clark, M. E. Jacka, W. A. Kukull et al., “The national alzheimer’s coordinating center (nacc) database: an alzheimer disease database,” Alzheimer Disease & Associated Disorders, vol. 18, no. 4, pp. 270–277, 2004

    2004

  29. [37]

    Ways toward an early diagnosis in alzheimer’s disease: the alzheimer’s disease neuroimaging initiative (adni),

    S. G. Mueller, M. W. Weiner, L. J. Thal, R. C. Petersen, C. R. Jack, W. Jagust, J. Q. Trojanowski, A. W. Toga, and L. Beckett, “Ways toward an early diagnosis in alzheimer’s disease: the alzheimer’s disease neuroimaging initiative (adni),” Alzheimer’s & Dementia, vol. 1, no. 1...

    2005

  30. [38]

    The australian biomarkers lifestyle and imaging flagship study of ageing,

    K. Ellis, D. Ames, R. Martins, P. Hudson, and C. Masters, “The australian biomarkers lifestyle and imaging flagship study of ageing,” Acta Neuropsychiatrica, vol. 18, no. 6, pp. 285–285, 2006

    2006

  31. [39]

    The parkinson progression marker initiative (ppmi),

    K. Marek, D. Jennings, S. Lasch, A. Siderowf, C. Tanner, T. Simuni, C. Coffey, K. Kieburtz, E. Flagg, S. Chowdhury et al., “The parkinson progression marker initiative (ppmi),” Progress in neurobiology, vol. 95, no. 4, pp. 629–635, 2011

    2011

  32. [40]

    Open access series of imaging studies: longitudinal mri data in nondemented and demented older adults,

    D. S. Marcus, A. F. Fotenos, J. G. Csernansky, J. C. Morris, and R. L. Buckner, “Open access series of imaging studies: longitudinal mri data in nondemented and demented older adults,” Journal of cognitive neuroscience, vol. 22, no. 12, pp. 2677–2684, 2010

    2010

  33. [41]

    Frontotemporal degeneration, the next therapeutic frontier: molecules and animal models for frontotemporal degeneration drug development,

    A. L. Boxer, M. Gold, E. Huey, F.-B. Gao, E. A. Burton, T. Chow, A. Kao, B. R. Leavitt, B. Lamb, M. Grether et al., “Frontotemporal degeneration, the next therapeutic frontier: molecules and animal models for frontotemporal degeneration drug development,” Alzheimer’s & Dementi...

    2013

  34. [42]

    Progression of brain atrophy in psp and cbs over 6 months and 1 year,

    S. Dutt, R. J. Binney, H. W. Heuer, P. Luong, S. Attygalle, P. Bhatt, G. A. Marx, J. Elofson, M. C. Tartaglia, I. Litvan et al., “Progression of brain atrophy in psp and cbs over 6 months and 1 year,” Neurology, vol. 87, no. 19, pp. 2016–2025, 2016

    2016

  35. [43]

    Large-scale long-tailed disease diagnosis on radiology images,

    Q. Zheng, W. Zhao, C. Wu, X. Zhang, L. Dai, H. Guan, Y . Li, Y . Zhang, Y . Wang, and W. Xie, “Large-scale long-tailed disease diagnosis on radiology images,” Nature Communications, vol. 15, no. 1, p. 10147, 2024

    2024

  36. [44]

    Multi-modal mamba modeling for survival prediction (m4survive): 14 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. XX, NO. XX, XXXX 2020 Adapting joint foundation model representations,

    H. H. Lee, A. Santamaria-Pang, J. Merkov, M. Lungren, and I. Tarapov, “Multi-modal mamba modeling for survival prediction (m4survive): 14 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. XX, NO. XX, XXXX 2020 Adapting joint foundation model representations,” arXiv preprint arXiv:250...

    2020 arXiv

  37. [45]

    Brainmvp: Multi-modal vision pre-training for brain image analysis using multi-parametric mri,

    S. Rui, L. Chen, Z. Tang, L. Wang, M. Liu, S. Zhang, and X. Wang, “Brainmvp: Multi-modal vision pre-training for brain image analysis using multi-parametric mri,” arXiv e-prints, pp. arXiv–2410, 2024

    2024

  38. [46]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  39. [47]

    Beta amyloid, tau, neuroimaging, and cognition: sequence modeling of biomarkers for alzheimer’s disease,

    S. D. Han, J. Gruhl, L. Beckett, H. H. Dodge, N. H. Stricker, S. Farias, D. Mungas, and A. D. N. Initiative, “Beta amyloid, tau, neuroimaging, and cognition: sequence modeling of biomarkers for alzheimer’s disease,” Brain imaging and behavior, vol. 6, no. 4, pp. 610–620, 2012

    2012

  40. [48]

    Depression case finding in individuals with dementia: a systematic review and meta-analysis,

    Z. S. Goodarzi, B. S. Mele, D. J. Roberts, and J. Holroyd-Leduc, “Depression case finding in individuals with dementia: a systematic review and meta-analysis,” Journal of the American Geriatrics Society, vol. 65, no. 5, pp. 937–948, 2017

    2017

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.