Pith. sign in

REVIEW 2 major objections 6 minor 242 references

Causal transfer learning frames medical domain shift as a causal problem so models keep the mechanisms that stay stable across hospitals and scanners.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-13 18:51 UTC pith:7LWQYYW5

load-bearing objection Solid survey that organises a messy literature under a CTL taxonomy; useful map, no new results, and the usual soft boundary between true causal ID and causality-inspired regularisers. the 2 major comments →

arxiv 2603.24388 v2 pith:7LWQYYW5 submitted 2026-03-25 cs.CV

Causal Transfer in Medical Image Analysis

Mohammed M. Abdelsamea , Daniel Tweneboah Anyimadu , Tasneem Selim , Saif Alzubi , Lei Zhang , Ahmed Karam Eldaly , Xujiong Ye This is my paper
classification cs.CV
keywords Causal Transfer Learningmedical image analysisdomain shiftinvariant risk minimisationstructural causal modelscounterfactual reasoningdomain adaptationclinical generalisation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Medical imaging AI often collapses when moved to a new hospital, scanner, or population because it has learned fragile correlations rather than disease mechanisms. This survey argues that the right fix is Causal Transfer Learning: embed structural causal models, invariant risk minimisation, and counterfactual reasoning inside ordinary transfer pipelines so that the features that matter for diagnosis remain invariant while scanner style and site artefacts are treated as non-causal. The authors organise more than eighty studies by task, type of shift, and causal assumption, supply a unified taxonomy, and collect the datasets and reported gains that show when the causal approach beats pure distribution alignment. A sympathetic reader cares because the same machinery is claimed to improve fairness, privacy-preserving federated learning, and clinical trust without requiring source data at deployment time.

Core claim

Domain shift in medical imaging is best understood as a violation of causal invariance; once structural causal models, invariant risk minimisation, and counterfactual reasoning are placed inside transfer-learning pipelines, the resulting Causal Transfer Learning recovers mechanisms that stay stable across hospitals, scanners, populations and protocols and therefore generalises more reliably than correlation-based domain adaptation.

What carries the argument

Causal Transfer Learning (CTL) — the joint embedding of structural causal models, invariant risk minimisation and counterfactual reasoning inside transfer pipelines — together with the three-axis taxonomy that organises methods by causal framework, causal operation, and role in the transfer pipeline.

Load-bearing premise

The surveyed methods recover genuine causal mechanisms that can be identified from ordinary observational medical images, rather than merely using causality-inspired regularisers or style-content tricks.

What would settle it

A controlled multi-hospital, multi-scanner benchmark in which a pure statistical domain-adaptation baseline matches or exceeds the cross-domain accuracy, fairness and robustness of the best CTL methods once both are given identical data and compute.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 6 minor

Summary. This survey introduces and systematises Causal Transfer Learning (CTL) for medical image analysis, framing domain shift as a causal problem and arguing that embedding structural causal models, invariant risk minimisation, and counterfactual reasoning inside transfer-learning pipelines yields invariant mechanisms that remain stable across hospitals, scanners, populations and protocols. It proposes a unified taxonomy (Figure 3) organising methods by causal framework, operation and transfer-pipeline role, reviews more than 80 studies across classification, segmentation, reconstruction, anomaly detection and multimodal tasks, summarises datasets and reported cross-domain gains, and discusses implications for fairness, federated settings and clinical trustworthiness. The central claim is organisational and synthetic rather than a new theorem or algorithm: CTL outperforms purely correlation-based domain adaptation for clinically reliable medical imaging AI.

Significance. If the organisational claim holds, the paper supplies the first comprehensive map that unifies causal inference with transfer learning specifically for medical imaging, filling a documented gap left by prior surveys that treat the two topics in isolation (Table 1). The taxonomy, task–shift–assumption tables (Tables 7–9) and curated dataset list (Table 10) give researchers a concrete reference for method selection and for identifying when causal assumptions are supported by evidence. Strengths include consistent numerical reporting of gains claimed by the original papers, explicit clinical-relevance subsections, and a clear statement of open challenges (scalability, validation, ethics, interpretability). Because the work is a literature synthesis rather than a derivation of new identification results, its lasting value lies in the taxonomy and the honest catalogue of remaining gaps rather than in any single empirical claim.

major comments (2)
  1. The manuscript repeatedly equates ‘causality-inspired’ regularisers and style–content disentanglements with recovery of identifiable causal mechanisms (abstract; §3; taxonomy of Figure 3; §6.1 CSSN Fourier style-swap; §6.3 GIN/IPA; §6.6 prototype-guided SFDA). Many of the >80 methods catalogued do not satisfy standard identification conditions from observational medical-image distributions. Because the paper never claims new identification theorems, this conflation does not invalidate the taxonomy, but it does over-state the central claim that CTL ‘identifies invariant mechanisms’. A short clarifying subsection (or revised wording throughout §3 and §6) that distinguishes true causal identification from causality-inspired regularisation is needed for the claim to remain proportionate.
  2. Section 10 and Table 11 list causal evaluation criteria (intervention testing, counterfactual reasoning, do-calculus validity, SHD, etc.), yet the empirical summaries in §6 report only conventional image-analysis metrics (Dice, AUROC, PSNR). The survey therefore never demonstrates that any of the reviewed methods actually satisfy the causal criteria it itself advocates. Either (a) extract and report any causal diagnostics present in the original papers, or (b) explicitly acknowledge that such diagnostics are almost never performed and treat this as a systematic gap. Without one of these steps the ‘when and why causal transfer outperforms’ claim remains under-supported.
minor comments (6)
  1. Abstract and §1 contain several grammatical slips (‘We studied spanning classification…’, ‘This is the generalisation problem…’) that should be corrected for readability.
  2. Figure 3 is referenced as the central taxonomy yet is never described in the main text beyond a caption; a short paragraph walking the reader through its four axes would improve accessibility.
  3. Table 1 comparison with prior surveys is useful, but the ‘Clinical Robustness’ column is binary and therefore uninformative; a brief qualitative note would strengthen the positioning claim.
  4. Equations (9)–(10) introduce IRM and the CTL objective without stating the precise environments or the form of the invariance penalty used by the surveyed medical-imaging papers; a forward pointer to the concrete realisations in §6 would help.
  5. Section 9 and Table 10 list many suitable datasets, yet several entries (e.g., IXI, fastMRI) lack explicit citation of the CTL papers that actually used them; adding those citations would close the loop.
  6. Occasional typographic inconsistencies appear (e.g., ‘Causal Treatment Learning (CTL)’ in §1 versus the rest of the paper; duplicated reference numbers for the same work). A final copy-edit pass is warranted.

Circularity Check

0 steps flagged

No significant circularity: survey taxonomy and literature synthesis do not derive predictions from their own inputs.

full rationale

This manuscript is a literature survey that defines Causal Transfer Learning (CTL), proposes an organisational taxonomy (Figure 3), and catalogues existing methods by task, shift type, and causal assumption. It does not fit parameters to data and then re-present those fits as predictions; it does not claim new identification theorems whose conclusions are built into the premises; and its central organisational claim is not load-bearing on self-citation uniqueness results. Reported empirical gains are attributed to the surveyed external studies (e.g., CSSN, GIN/IPA, CauSSL, GenCA-MRI), not derived by construction from the survey’s own definitions. Self-citations (e.g., Capri-CT, authors’ related work) appear only as peripheral application examples and do not close any logical loop. Renaming and grouping prior causality-inspired transfer methods under the CTL label is standard survey practice, not a circular derivation. Score 0 is therefore appropriate.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 1 invented entities

A survey inherits the standard assumptions of causal inference and transfer learning; it introduces no free parameters and only one named paradigm (CTL). The load-bearing background claims are the usual causal-identification conditions and the empirical reports of the cited papers.

axioms (4)
  • domain assumption Ignorability / no unobserved confounding given observed covariates (Eq. 1)
    Invoked throughout Sections 2–3 to justify that causal features can be recovered from medical images.
  • domain assumption Existence of invariant causal mechanisms across environments (IRM premise)
    Core of the CTL objective (Eq. 9–10) and of the taxonomy’s ‘invariance’ branch.
  • domain assumption Structural causal models correctly capture the generative process of medical images (style vs content, scanner as intervention)
    Used to interpret Fourier style transfer, GIN/IPA augmentations and GenCA-MRI as causal interventions.
  • standard math Standard do-calculus and potential-outcomes frameworks are valid for imaging interventions
    Background machinery of Pearl and Rubin cited in Section 2.
invented entities (1)
  • Causal Transfer Learning (CTL) paradigm independent evidence
    purpose: Umbrella term that unifies SCMs, IRM and counterfactual methods inside transfer-learning pipelines for medical imaging.
    The paper coins and systematises the label; independent evidence consists of the performance gains reported by the individual methods it groups under the label.

pith-pipeline@v1.1.0-grok45 · 50252 in / 2284 out tokens · 36571 ms · 2026-07-13T18:51:47.339915+00:00 · methodology

0 comments
read the original abstract

Medical imaging models frequently fail when deployed across hospitals, scanners, populations, or imaging protocols due to domain shift, limiting their clinical reliability. While transfer learning and domain adaptation address such shifts statistically, they often rely on spurious correlations that break under changing conditions. On the other hand, causal inference provides a principled way to identify invariant mechanisms that remain stable across environments. This survey introduces and systematises Causal Transfer Learning (CTL) for medical image analysis. This paradigm integrates causal reasoning with cross-domain representation learning to enable robust and generalisable clinical AI. We frame domain shift as a causal problem and analyse how structural causal models, invariant risk minimisation, and counterfactual reasoning can be embedded within transfer learning pipelines. We studied spanning classification, segmentation, reconstruction, anomaly detection, and multimodal imaging, and organised them by task, shift type, and causal assumption. A unified taxonomy is proposed that connects causal frameworks and transfer mechanisms. We further summarise datasets, benchmarks, and empirical gains, highlighting when and why causal transfer outperforms correlation-based domain adaptation. Finally, we discuss how CTL supports fairness, robustness, and trustworthy deployment in multi-institutional and federated settings, and outline open challenges and research directions for clinically reliable medical imaging AI.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

242 extracted references · 1 canonical work pages

  1. [1]

    Litjens, T

    G. Litjens, T. Kooi, B. E. Bejnordi, A. A. A. Setio, F. Ciompi, M. Ghafoorian, J. A. van der Laak, B. van Ginneken, C. I. Sanchez, A survey on deep learning in medical image analysis, Medical image analysis 42 (2017) 60–88

    2017

  2. [2]

    LeCun, Y

    Y. LeCun, Y. Bengio, G. Hinton, Deep learning, nature 521 (7553) (2015) 436–444

    2015

  3. [3]

    J. A. Cortes-Briones, N. I. Tapia-Rivas, D. C. D’Souza, P. A. Estevez, Going deep into schizophrenia with artificial intelligence, Schizophrenia Research (2021)

    2021

  4. [4]

    Gulrajani, D

    I. Gulrajani, D. Lopez-Paz, In search of lost domain generalization, arXiv preprint arXiv:2007.01434 (2021)

    Pith/arXiv arXiv 2007

  5. [5]

    Subbaswamy, R

    A. Subbaswamy, R. Adams, S. Saria, Evaluating model robustness and stability to dataset shift, in: International conference on artificial intelligence and statistics, PMLR, 2021, pp. 2611–2619

    2021

  6. [6]

    Bernhardt, C

    M. Bernhardt, C. Jones, B. Glocker, Investigating underdiagnosis of ai algorithms in the presence of multiple sources of dataset bias, arXiv preprint arXiv:2201.07856 (2022)

    Pith/arXiv arXiv 2022

  7. [7]

    R. Wang, P. Chaudhari, C. Davatzikos, Harmonization with flow-based causal inference, in: M. de Bruijne, P. C. Cattin, S. Cotin, N. Padoy, S. Speidel, Y. Zheng, C. Essert (Eds.), Medi- cal Image Computing and Computer Assisted Intervention–MICCAI 2021, Springer International Publishing, Cham, 2021, pp. 181–190

    2021

  8. [8]

    H. Ye, C. Xie, Y. Liu, Z. Li, Out-of-distribution generalization analysis via influence function, arXiv preprint arXiv:2101.08521 (2021)

    Pith/arXiv arXiv 2021

  9. [9]

    Zhang, N

    H. Zhang, N. Dullerud, L. Seyyed-Kalantari, Q. Morris, S. Joshi, M. Ghassemi, An empirical framework for domain generalization in clinical settings, in: Proceedings of the Conference on Health, Inference, and Learning (CHIL ’21), Association for Computing Machinery, New York, NY, USA, 2021, pp. 279–290

    2021

  10. [10]

    Valvano, A

    G. Valvano, A. Leo, S. A. Tsaftaris, Re-using adversarial mask discriminators for test-time train- ing under distribution shifts, Machine Learning for Biomedical Imaging, MICCAI 2021 workshop omnibus special issue (2021)

    2021

  11. [11]

    Vlontzos, G

    A. Vlontzos, G. Sutherland, S. Ganju, F. Soboczenski, Next-gen machine learning supported diag- nostic systems for spacecraft, in: AI for Spacecraft Longevity Workshop at IJCAI, 2021

    2021

  12. [12]

    Prosperi, Y

    M. Prosperi, Y. Guo, M. Sperrin, J. S. Koopman, J.-S. Min, X. He, S. Rich, M. Wang, I. E. Buchan, J. Bian, Causal inference and counterfactual prediction in machine learning for actionable healthcare, Nature Machine Intelligence 2 (7) (2020) 369–375

    2020

  13. [13]

    Hirano, A

    H. Hirano, A. Minagi, K. Takemoto, Universal adversarial attacks on deep neural networks for medical image classification, BMC Medical Imaging 21 (1) (2021) 1–13. 35

    2021

  14. [14]

    Huang, K

    B. Huang, K. Zhang, J. Zhang, J. Ramsey, R. Sanchez-Romero, C. Glymour, B. Schölkopf, Causal discovery from heterogeneous/nonstationary data with independent changes, Journal of Machine Learning Research 21 (89) (2020) 1–53

    2020

  15. [15]

    Kayser, R

    M. Kayser, R. D. Soberanis-Mukul, A.-M. Zvereva, P. Klare, N. Navab, S. Albarqouni, Under- standing the effects of artifacts on automated polyp detection and incorporating that knowledge via learning without forgetting, arXiv preprint arXiv:2002.02883 (2020)

    Pith/arXiv arXiv 2002

  16. [16]

    Lavin, C

    A. Lavin, C. M. Gilligan-Lee, A. Visnjic, S. Ganju, D. Newman, S. Ganguly, D. Lange, A. G. Baydin, A. Sharma, A. Gibson, et al., Technology readiness levels for machine learning systems, arXiv preprint arXiv:2101.03989 (2021)

    Pith/arXiv arXiv 2021

  17. [17]

    P. M. Gordaliza, J. J. Vaquero, A. Munoz-Barrutia, Translational lung imaging analysis through disentangled representations, arXiv preprint arXiv:2203.01668 (2022)

    Pith/arXiv arXiv 2022

  18. [18]

    Grzech, B

    D. Grzech, B. Kainz, B. Glocker, L. Le Folgoc, Image registration via stochastic gradient markov chain monte carlo, in: International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, Springer, 2020, pp. 3–12

    2020

  19. [19]

    Y.Zhang, M.Gong, T.Liu, G.Niu, X.Tian, B.Han, B.Schölkopf, K.Zhang, Adversarialrobustness through the lens of causality, arXiv preprint arXiv:2106.06196 (2022)

    Pith/arXiv arXiv 2022

  20. [20]

    B. G. Santa Cruz, C. Vega, F. Hertel, The need of standardised metadata to encode causal relation- ships: Towards safer data-driven machine learning biological solutions, in: Proceedings of CIBB, 2021, p. 1

    2021

  21. [21]

    S. J. Pan, Q. Yang, A survey on transfer learning, IEEE Transactions on knowledge and data engineering 22 (10) (2010) 1345–1359

    2010

  22. [22]

    Pearl, Causality, Cambridge university press, 2009

    J. Pearl, Causality, Cambridge university press, 2009

    2009

  23. [23]

    Schölkopf, F

    B. Schölkopf, F. Locatello, S. Bauer, N. R. Ke, N. Kalchbrenner, A. Goyal, Y. Bengio, Toward causal representation learning, Proceedings of the IEEE 109 (5) (2021) 612–634

    2021

  24. [24]

    K. Yi, C. Gan, Y. Li, P. Kohli, J. Wu, A. Torralba, J. B. Tenenbaum, Clevrer: Collision events for video representation and reasoning, in: International Conference on Learning Representations, 2020

    2020

  25. [25]

    Benkarim, C

    O. Benkarim, C. Paquola, B.-Y. Park, V. Kebets, S.-J. Hong, R. V. de Wael, S. Zhang, B. T. T. Yeo, M. Eickenberg, T. Ge, et al., The cost of untracked diversity in brain-imaging prediction, bioRxiv (2021)

    2021

  26. [26]

    S.Budd, E.C.Robinson, B.Kainz, Asurveyonactivelearningandhuman-in-the-loopdeeplearning for medical image analysis, Medical Image Analysis 71 (2021) 102062

    2021

  27. [27]

    Schrouff, N

    J. Schrouff, N. Harris, O. Koyejo, I. Alabdulmohsin, E. Schnider, K. Opsahl-Ong, A. Brown, S. Roy, D. Mincu, C. Chen, et al., Maintaining fairness across distribution shift: Do we have viable solutions for real-world applications?, arXiv preprint arXiv:2202.01034 (2022)

    Pith/arXiv arXiv 2022

  28. [28]

    Singla, S

    S. Singla, S. Wallace, S. Triantafillou, K. Batmanghelich, Using causal analysis for conceptual deep learning explanation, in: M. de Bruijne, P. C. Cattin, S. Cotin, N. Padoy, S. Speidel, Y. Zheng, C. Essert (Eds.), Medical Image Computing and Computer Assisted Intervention–MICCAI 2021, Springer International Publishing, Cham, 2021, pp. 519–528

    2021

  29. [29]

    Ouyang, C

    C. Ouyang, C. Chen, S. Li, Z. Li, C. Qin, W. Bai, D. Rueckert, Causality-inspired single-source domain generalization for medical image segmentation, arXiv preprint arXiv:2111.12525 (2021)

    Pith/arXiv arXiv 2021

  30. [30]

    S. Li, M. Sesia, Y. Romano, E. Candès, C. Sabatti, Searching for consistent associations with a multi-environment knockoff filter, arXiv preprint arXiv:2106.04118 (2021). 36

    Pith/arXiv arXiv 2021

  31. [31]

    Chuang, S

    K.-C. Chuang, S. Ramakrishnapillai, L. Bazzano, O. Carmichael, Nonlinear conditional time- varying granger causality of task fmri via deep stacking networks and adaptive convolutional ker- nels, in: L. Wang, Q. Dou, P. T. Fletcher, S. Speidel, S. Li (Eds.), Medical Image Computing and Computer Assisted Intervention – MICCAI 2022, Springer Nature Switzerla...

    2022

  32. [32]

    H. Ding, J. Zhang, P. Kazanzides, J. Y. Wu, M. Unberath, Carts: Causality-driven robot tool segmentation from vision and kinematics data, in: L. Wang, Q. Dou, P. T. Fletcher, S. Speidel, S. Li (Eds.), Medical Image Computing and Computer Assisted Intervention – MICCAI 2022, Springer Nature Switzerland, Cham, 2022, pp. 387–398

    2022

  33. [33]

    Adebayo, M

    J. Adebayo, M. Muelly, H. Abelson, B. Kim, Post hoc explanations may be ineffective for detecting unknown spurious correlation, in: International Conference on Learning Representations, 2022. URLhttps://openreview.net/forum?id=xNOVfCCvDpM

    2022

  34. [34]

    S. Mani, G. F. Cooper, Causal discovery from medical textual data, in: Proceedings of the AMIA Symposium, American Medical Informatics Association, 2000, p. 542

    2000

  35. [35]

    Pölsterl, C

    S. Pölsterl, C. Wachinger, Estimation of causal effects in the presence of unobserved confounding in the alzheimer’s continuum, in: Information Processing in Medical Imaging, Springer, Cham, 2021, pp. 45–57

    2021

  36. [36]

    J. D. Ramsey, S. J. Hanson, C. Hanson, Y. O. Halchenko, R. A. Poldrack, C. Glymour, Six problems for causal inference from fmri, NeuroImage 49 (2) (2010) 1545–1558

    2010

  37. [37]

    N. R. Ke, S. Chiappa, J. Wang, J. Bornschein, T. Weber, A. Goyal, M. Botvinick, M. Mozer, D. J. Rezende, Learning to induce causal structure, arXiv preprint arXiv:2203.01774 (2022)

    Pith/arXiv arXiv 2022

  38. [38]

    Clivio, F

    O. Clivio, F. Falck, B. Lehmann, G. Deligiannidis, C. Holmes, Neural score matching for high- dimensional causal inference, in: AISTATS, 2022

    2022

  39. [39]

    Uhler, J

    C. Uhler, J. Zhang, Causal structure and representation learning with biomedical applications, arXiv preprint arXiv:2511.04790 (2025)

    arXiv 2025

  40. [40]

    G.Carloni, Human-aligneddeeplearning: Explainability, causality, andbiologicalinspiration, arXiv preprint arXiv:2504.13717 (2025)

    Pith/arXiv arXiv 2025

  41. [41]

    Mesinovic, M

    M. Mesinovic, M. Buhlan, T. Zhu, Causal graph neural networks for healthcare, arXiv preprint arXiv:2511.02531 (2025)

    arXiv 2025

  42. [42]

    J. Fehr, M. Piccininni, T. Kurth, S. Konigorski, A causal framework for assessing the transporta- bility of clinical prediction models, medRxiv (2022)

    2022

  43. [43]

    L. Fay, H. Reguigui, B. Yang, S. Gatidis, T. Küstner, Mimm-x: Disentangling spurious correlations for medical image analysis, in: MICCAI Workshop on Fairness of AI in Medical Imaging, Springer, 2025, pp. 94–103

    2025

  44. [44]

    Rojas-Carulla, B

    M. Rojas-Carulla, B. Schölkopf, R. Turner, J. Peters, Invariant models for causal transfer learning, Journal of Machine Learning Research 19 (36) (2018) 1–34

    2018

  45. [45]

    Zapaishchykova, D

    A. Zapaishchykova, D. Dreizin, Z. Li, J. Y. Wu, S. Faghihroohi, M. Unberath, An interpretable approach to automated severity scoring in pelvic trauma, in: Medical Image Computing and Computer-Assisted Intervention (MICCAI), Springer, 2021, pp. 424–433

    2021

  46. [46]

    Hussain, F

    M. Hussain, F. A. Satti, J. Hussain, T. Ali, S. I. Ali, H. S. M. Bilal, G. H. Park, S. Lee, T. Chung, A practical approach towards causality mining in clinical text using active transfer learning, Journal of Biomedical Informatics 123 (2021) 103932

    2021

  47. [47]

    C. Liu, X. Sun, J. Wang, H. Tang, T. Li, T. Qin, W. Chen, T.-Y. Liu, Learning causal semantic representation for out-of-distribution prediction, in: Advances in Neural Information Processing Systems, Vol. 34, 2021. 37

    2021

  48. [48]

    Fawkes, R

    J. Fawkes, R. Evans, D. Sejdinovic, Selection, ignorability and challenges with causal fairness, in: Conference on Causal Learning and Reasoning, PMLR, 2022, pp. 275–289

    2022

  49. [49]

    D. B. Rubin, Bayesian inference for causal effects: The role of randomization, The Annals of Statistics 6 (1) (1978) 34–58

    1978

  50. [50]

    Vlontzos, D

    A. Vlontzos, D. Rueckert, B. Kainz, A review of causality for learning algorithms in medical image analysis, arXiv preprint arXiv:2206.05498 (2022)

    Pith/arXiv arXiv 2022

  51. [51]

    L. G. Neuberg, Causality: models, reasoning, and inference, by judea pearl, cambridge university press, 2000, Econometric Theory 19 (4) (2003) 675–685

    2000

  52. [52]

    F. Boge, A. Mosig, Causality and scientific explanation of artificial intelligence systems in biomedicine, Pflügers Archiv-European Journal of Physiology 477 (4) (2025) 543–554

    2025

  53. [53]

    P. C. Austin, An introduction to propensity score methods for reducing the effects of confounding in observational studies, Multivariate behavioral research 46 (3) (2011) 399–424

    2011

  54. [54]

    A. J. Sedgewick, K. Buschur, I. Shi, J. D. Ramsey, V. K. Raghu, D. V. Manatakis, Y. Zhang, J. Bon, D. Chandra, C. Karoleski, et al., Mixed graphical models for integrative causal analysis with application to chronic lung disease diagnosis and prognosis, Bioinformatics 35 (7) (2019) 1204– 1212

    2019

  55. [55]

    L. Yao, Z. Chu, S. Li, Y. Li, J. Gao, A. Zhang, A survey on causal inference, ACM Transactions on Knowledge Discovery from Data (TKDD) 15 (5) (2021) 1–46

    2021

  56. [56]

    Ghosh, E

    D. Ghosh, E. Mastej, R. Jain, Y. S. Choi, Causal inference in radiomics: Framework, mechanisms, and algorithms, Frontiers in Neuroscience 16 (2022) 884708

    2022

  57. [57]

    J. G. Richens, C. M. Lee, S. Johri, Improving the accuracy of medical diagnosis with causal machine learning, Nature communications 11 (1) (2020) 3923

    2020

  58. [58]

    Nauta, D

    M. Nauta, D. Bucur, C. Seifert, Causal discovery with attention-based convolutional neural net- works, Machine Learning and Knowledge Extraction 1 (1) (2019) 312–340

    2019

  59. [59]

    Gerstenberg, N

    T. Gerstenberg, N. D. Goodman, D. A. Lagnado, J. B. Tenenbaum, A counterfactual simulation model of causal judgments for physical events, Psychological Review 128 (5) (2021) 936

    2021

  60. [60]

    Sanchez, J

    P. Sanchez, J. P. Voisey, T. Xia, H. I. Watson, A. Q. O’Neil, S. A. Tsaftaris, Causal machine learning for healthcare and precision medicine, Royal Society Open Science 9 (8) (2022) 220638

    2022

  61. [61]

    D. B. Rubin, Estimating causal effects of treatments in randomized and nonrandomized studies., Journal of educational Psychology 66 (5) (1974) 688

    1974

  62. [62]

    G. W. Imbens, D. B. Rubin, Causal inference in statistics, social, and biomedical sciences, Cam- bridge university press, 2015

    2015

  63. [63]

    P. R. Rosenbaum, D. B. Rubin, The central role of the propensity score in observational studies for causal effects, Biometrika 70 (1) (1983) 41–55

    1983

  64. [64]

    Gelman, J

    A. Gelman, J. B. Carlin, H. S. Stern, D. B. Dunson, A. Vehtari, D. B. Rubin, Bayesian Data Analysis, 3rd Edition, Chapman and Hall/CRC, Boca Raton, FL, 2013

    2013

  65. [65]

    D. P. Kingma, M. Welling, Auto-encoding variational bayes, arXiv preprint arXiv:1312.6114 (2013)

    Pith/arXiv arXiv 2013

  66. [66]

    Pearl, Causality: Models, Reasoning, and Inference, Cambridge University Press, 2000

    J. Pearl, Causality: Models, Reasoning, and Inference, Cambridge University Press, 2000

    2000

  67. [67]

    D. B. Rubin, Causal inference using potential outcomes: Design, modeling, decisions, Journal of the American statistical Association 100 (469) (2005) 322–331

    2005

  68. [68]

    Ibeling, T

    D. Ibeling, T. Icard, Comparing causal frameworks: Potential outcomes, structural models, graphs, and abstractions, Advances in Neural Information Processing Systems 36 (2023) 80130–80141. 38

    2023

  69. [69]

    X. Wu, S. Peng, J. Li, J. Zhang, Q. Sun, W. Li, Q. Qian, Y. Liu, Y. Guo, Causal inference in the medical domain: A survey, Applied Intelligence 54 (6) (2024) 4911–4934

    2024

  70. [70]

    D. C. Castro, I. Walker, B. Glocker, Causality matters in medical imaging, Nature Communications 11 (1) (2020) 3673

    2020

  71. [71]

    Petersen, E

    E. Petersen, E. Ferrante, M. Ganz, A. Feragen, Are demographically invariant models and repre- sentations in medical imaging fair?, arXiv preprint arXiv:2305.01397 (2023)

    Pith/arXiv arXiv 2023

  72. [72]

    Papangelou, K

    K. Papangelou, K. Sechidis, J. Weatherall, G. Brown, Toward an understanding of adversarial examples in clinical trials, in: Joint European Conference on Machine Learning and Knowledge Discovery in Databases, Springer, 2018, pp. 35–51

    2018

  73. [73]

    Huang, T

    Y. Huang, T. Würfl, K. Breininger, L. Liu, G. Lauritsch, A. Maier, Abstract: Some investigations on robustness of deep learning in limited angle tomography, in: MICCAI, Springer, 2019, pp. 21–21

    2019

  74. [74]

    A. J. DeGrave, J. D. Janizek, S.-I. Lee, Ai for radiographic covid-19 detection selects shortcuts over signal, Nature Machine Intelligence (2021)

    2021

  75. [75]

    B. G. S. Cruz, A. Husch, F. Hertel, The effect of dataset confounding on predictions of deep neural networks for medical imaging, arXiv preprint (2021)

    2021

  76. [76]

    T. Chen, S. Kornblith, M. Noroozi, A. Hwang, A simple framework for contrastive learning of visual representations, arXiv preprint arXiv:2002.05709 (2020)

    Pith/arXiv arXiv 2002

  77. [77]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, et al., Learning transferable visual models from natural language supervi- sion, in: International conference on machine learning, PmLR, 2021, pp. 8748–8763

    2021

  78. [78]

    X. Zhai, B. Mustafa, A. Kolesnikov, L. Beyer, Sigmoid loss for language image pre-training, in: Proceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 11975–11986

    2023

  79. [79]

    Beyer, Y

    M.Tschannen, A.Gritsenko, X.Wang, M.F.Naeem, I.Alabdulmohsin, N.Parthasarathy, T.Evans, L. Beyer, Y. Xia, B. Mustafa, et al., Siglip 2: Multilingual vision-language encoders with improved semantic understanding, localization, and dense features, arXiv preprint arXiv:2502.14786 (2025)

    Pith/arXiv arXiv 2025

  80. [80]

    Zhang, F

    H. Zhang, F. Li, S. Liu, L. Zhang, H. Su, J. Zhu, L. M. Ni, H.-Y. Shum, Dino: Detr with improved denoising anchor boxes for end-to-end object detection, arXiv preprint arXiv:2203.03605 (2022)

    Pith/arXiv arXiv 2022

Showing first 80 references.