Pith. sign in

REVIEW 3 major objections 5 minor 27 references

Bayesian uncertainty estimation improves clinical decision making in medical AI agents

T0 review · 3 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read A chest-radiograph AI classifier's own uncertainty, delivered to a downstream agent as a binary error-risk flag rather than raw numbers, lets the agent catch the model's errors at the achievable optimum and cuts confident misdiagnoses on un

desk verdict The flag-vs-raw agent result is confounded by the fitted risk score, but the uncertainty error-detection signal is real and worth a closer look. read the letter →

arxiv 2607.20582 v1 pith:CUT6QMA6 submitted 2026-07-22 cs.LG cs.AIcs.MA

Frederik Hauke , Patrick Wienholt , Christiane Kuhl , Dyke Ferber , Jakob Nikolas Kather , Sven Nebelung , Daniel Truhn This is my paper · ORCID
classification cs.LGcs.AIcs.MA
keywords MonteCarlodropoutepistemicuncertaintyclinicaldecisionsupportchestradiographymedicalAIagentserrordetectionrepresentationvisiontransformer
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 tries to establish that Monte Carlo dropout provides a usable epistemic uncertainty signal for a medical image classifier, and that the way this uncertainty is communicated to a downstream AI agent determines whether it improves decisions. In a controlled factorial experiment, the same uncertainty information helped a clinical-decision-support agent reach the model's error-detection frontier only when given as a binary error-risk flag, not as raw scores. The authors argue this is the first controlled demonstration that representation of uncertainty, not merely its availability, governs its value in clinical AI. The result matters because medical AI agents are increasingly being placed downstream of quantitative predictors, and it suggests a concrete design principle: keep numerical reasoning in the specialist model and hand the agent a formalized decision.

What carries the argument

The key mechanism is Monte Carlo dropout applied at inference time to a vision-transformer classifier: dropout layers remain active, and 30 stochastic forward passes produce a per-class predictive standard deviation, the epistemic uncertainty estimate. This standard deviation is then fed into a logistic-regression error detector fitted on the validation set and thresholded at its Youden-optimal point, yielding a binary error-risk flag (elevated risk vs normal). The flag does the load-bearing work: it converts continuous uncertainty into a decision the agent can follow, whereas raw numbers demand interpretation that the tested agent fails to perform optimally.

What would settle it

Run the same 2x2 factorial with a different language-model agent (or with clinician assessors reading the same outputs) and check whether the raw-number condition still underperforms the flag; also test a flag thresholded on an independent calibration set rather than the same detector used to define the ROC frontier.

Watch

Extended reading notes

Core claim

On a multi-task chest-radiograph classifier with eight thoracic findings, the authors show that predictive standard deviation from 30 Monte Carlo dropout forward passes tracks validation loss across training-set sizes and flags individual predictions that are confident yet wrong. Adding this signal to the point prediction raises error-detection AUROC from 0.74 to 0.77. But when a clinical-decision-support agent received the identical signal as raw numbers, it operated below the achievable frontier; when the same signal was pre-digested into a binary error-risk flag, the agent reached the frontier and cut confident misdiagnoses on unreliable findings from 8.5% to 2.7%. The conclusion is that

Load-bearing premise

The central conclusion rests on taking the language-model agent's structured commit/escalate outputs as valid clinical decisions, and the authors note these were not adjudicated by clinicians; if a different agent model or prompt can handle raw numbers well, the raw-versus-flag gap could shrink or vanish.

Editorial extensions

If this is right

  • Clinical AI systems should expose pre-digested, calibrated uncertainty flags to downstream agents rather than raw model outputs, if the observed effect generalizes.
  • MC dropout offers a practical, inference-time-only way to obtain confidence signals that track generalisation, requiring no retraining of separate ensembles.
  • The flag's selective escalation could order confirmatory tests (e.g., radiologist second read, CT) on unreliable findings while leaving reliable calls untouched, potentially improving workflow efficiency.
  • The design principle extends beyond uncertainty: any continuous model output (e.g., risk scores) may need formalization into discrete decisions before an agent can use it effectively.

Reading between the lines

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

  • The raw-versus-flag gap may be partly model-specific or prompt-sensitive; a different agent model or a differently phrased prompt might interpret raw uncertainty numbers correctly, so the design principle may not be universal.
  • The binary flag is constructed from the same logistic error detector used to draw the ROC frontier, so the 'flag reaches the frontier' result is partly engineered; a fairer test would calibrate the flag on a separate dataset.
  • The agent's commit/escalate decisions were not adjudicated by clinicians; a real clinical trial could reveal whether the flag's benefit survives human oversight and whether it changes workflow in unexpected ways.
  • The absolute error-detection gain from adding uncertainty (+0.023 AUROC) is modest; the representation effect is larger, suggesting that how uncertainty is communicated may matter more than how it is estimated.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper combines a mechanistic analysis of MC-dropout uncertainty in a DINOv2 chest-radiograph classifier with a GPT-5.1 agent experiment. Across training-set scales, predictive standard deviation tracks validation loss; per-class scatter plots show error enrichment at high standard deviation. On the held-out test set, a logistic-regression score combining confidence and uncertainty raises error-detection AUROC from 0.74 to 0.77. In a 2x2 factorial design, a binary error-risk flag leads the agent to escalate more on unreliable findings and reduces confident misdiagnoses from 8.5% to 2.7%, whereas raw numeric outputs do not. The authors conclude that representation, not just availability, determines whether uncertainty benefits downstream agents.

Significance. The empirical phenomenon—agents using a precomputed flag better than raw numbers—is practically important for clinical AI design. The paper has several strengths: held-out test evaluation, case-clustered bootstrap CIs, open code and data, per-class reliability analysis, and explicit limitations. However, the central flag-vs-raw comparison is confounded by the flag being a thresholded version of the same fitted logistic-regression risk score that defines the ROC frontier, and the agent evaluation is based on one LLM without clinician adjudication. These issues require additional controls or substantially weakened claims.

major comments (3)
  1. [Methods 3.5, Figure 5a, §1.5] The central comparison is confounded. The binary flag is produced by a logistic-regression error detector (on standardized confidence and uncertainty) fitted on the validation set and thresholded at its Youden-optimal point, while the prediction+uncertainty ROC frontier is the out-of-sample probability of a 5-fold cross-validated logistic regression on the same standardized inputs. Thus the flag condition receives a fitted, calibrated risk score, whereas the raw condition receives only the component inputs (prediction, MC-dropout standard deviation, orientation statistics). The comparison therefore varies both binarization and the presence of the precomputed risk score. The statement in §1.5 that 'the identical error-relevant signal was present in both representations' is misleading. A control arm in which the agent receives the continuous logistic-regression risk score is needed to isol
  2. [Figure 5a, §1.4] The headline raw-vs-flag gain is reported as a difference in sensitivity (+0.16) without the specificities at which these sensitivities are achieved. Operating points on an ROC curve trade sensitivity against specificity, so comparing sensitivity alone is not meaningful unless specificity is matched. Please report the full (sensitivity, specificity) coordinates for all four conditions and compare at matched specificity, or use a scalar such as partial AUC or net benefit. The statement that the raw point is 'well below the achievable frontier' should be explicitly evaluated at the raw point's own specificity.
  3. [Methods 3.5, Limitations, Abstract/Discussion] All agent-level conclusions rest on a single LLM (GPT-5.1) reading structured fields without clinician adjudication. The authors disclose this in the Limitations, but the abstract and Discussion make general claims about 'medical AI agents'. The raw-vs-flag gap could be model-specific or prompt-sensitive. Either add robustness checks with at least one additional LLM and/or prompt variation, or restrict the conclusions to the specific GPT-5.1 agent used here.
minor comments (5)
  1. [Affiliations] 'University Hospital R WTH Aachen' should be 'RWTH Aachen'.
  2. [Figure 2 caption] 'T raining fit' contains a stray space; should be 'Training fit'.
  3. [Methods 3.5] Please specify on which sample the 5-fold cross-validated logistic regression for the ROC frontier is computed (the 500-case test subset, presumably) and whether the folds are case-level, to avoid ambiguity.
  4. [Methods 3.5] Define 'Youden-optimal point' explicitly (the threshold maximizing sensitivity + specificity − 1) for readers unfamiliar with the term.
  5. [Figure 4] The scatter plots have substantial overplotting; consider density contours or alpha-blending to make the reported enrichment visible.

Circularity Check

1 steps flagged · score 6.0 of 10

Flag-vs-raw comparison is confounded: the flag is a thresholded logistic-regression detector and the frontier is the ROC of the same detector, so reaching the frontier is by construction and representation is not isolated.

  1. fitted input called prediction [Methods §3.5; Results §1.4/Figure 5a]
    "fitted on the validation set and thresholded at its Youden-optimal operating point // The X-Ray model’s error-detection frontiers (Figure 5a) are the ROC of two scores ... prediction plus uncertainty (the out-of-sample probability of a 5-fold cross-validated logistic regression on the standardised confidence and uncertainty)."

    The flag is a thresholded version of a validation-fitted logistic-regression error detector, and the frontier is the out-of-sample ROC of a 5-fold cross-validated logistic regression on the same standardized confidence and uncertainty features. Since the flag is constructed as the Youden-optimal operating point of that detector, following the flag places the agent on the detector's ROC by construction; 'reaching the frontier' is thus engineered, not discovered. The raw condition, in contrast, receives only the component inputs (deterministic prediction, MC-dropout standard deviation, orientation quantiles) and not the fitted logistic score, so the 2x2 varies both representation and the presence of a precomputed risk score. The Discussion's claim that 'the identical error-relevant signal wa

full rationale

The MC-dropout signal tracking validation loss (Figs 2-3), the reliability-plane enrichment (Fig 4), and the held-out AUROC gain from 0.74 to 0.77 are not circular: they are data-driven and benchmarked on held-out data. The circularity lies in the factorial agent comparison. The binary flag is produced by thresholding a validation-fitted logistic-regression error detector at its Youden-optimal point, while the reference boundary is the ROC of a logistic-regression error detector on the same features. The flag is therefore designed to be the optimal operating point of the detector whose ROC is used as the frontier; a compliant agent that follows the flag will sit on that ROC by construction. Additionally, the raw arm lacks the logistic-regression output entirely and receives only the raw confidence, MC standard deviation and orientation quantiles. Thus the experiment does not isolate binary-vs-continuous representation; it also changes whether a precomputed calibrated risk score is provided. The paper's assertion that 'the identical error-relevant signal was present in both representations' is contradicted by the Methods. This partial circularity affects the paper's central representational claim, though the underlying uncertainty signal retains independent empirical support.

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

No new physical or model-level entities are introduced. The free parameters are the validation-fitted error detector and its threshold, plus the manually chosen MC-dropout settings. The main domain assumptions are the MC-dropout epistemic interpretation, the binarization of equivocal grades, and the validity of GPT-5.1 structured outputs as clinical decisions.

free parameters (3)
  • Youden-optimal threshold for binary error-risk flag = not reported
    Fitted on the validation set (Methods §3.5). The flag agent's location on the ROC frontier is determined by this threshold, so part of the claimed flag-agent improvement is inherited from fitting rather than from the agent's uncertainty reasoning.
  • Logistic-regression error-detector coefficients (confidence and uncertainty) = not reported
    Fitted on validation (and in 5-fold CV for the test frontiers) to combine confidence and MC standard deviation. The AUROC 0.74-to-0.77 difference is measured on this detector, not on a parameter-free rule.
  • MC dropout rate and number of forward passes = p=0.1, 30 passes
    Chosen by hand without reported sensitivity analysis (Methods §3.2). The entire epistemic uncertainty signal depends on these values, and no alternative settings are shown.
assumptions (4)
  • domain assumption MC-dropout predictive standard deviation approximates epistemic uncertainty in the fine-tuned ViT.
    The paper takes predictive std dev as the epistemic uncertainty signal without validating the approximation against true model misspecification (Introduction, Methods §3.2).
  • domain assumption Binarizing ordinal grades at threshold >0 preserves clinically meaningful error labels.
    Grade 1 ('(+)', Borderline) is treated as positive in training and error scoring. The effect of this equivocal-label choice on the error-detection and misdiagnosis metrics is not analyzed (Methods §3.1).
  • domain assumption GPT-5.1 structured commit/escalate outputs can be read as simulated clinical decisions.
    The authors state decisions were read from a structured field the agent emitted and were not adjudicated by clinician assessors. All agent-level conclusions rest on this assumption.
  • standard math Standard statistical machinery for ROC, logistic regression, and case-clustered bootstrap.
    These are conventional methods applied without derivation; they are sufficiently standard to use as background.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bayesian uncertainty estimation improves clinical decision making in medical AI agents." pith.science (2026). https://pith.science/paper/CUT6QMA6

@misc{pith2026260720582,
  author       = {Pith},
  title        = {Pith review of: Bayesian uncertainty estimation improves clinical decision making in medical AI agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CUT6QMA6}},
  note         = {Machine review of arXiv:2607.20582}
}
abstract

Machine learning models for medical image analysis typically lack a reliable measure of confidence, limiting their use in ambiguous or atypical cases. Here we show that Monte Carlo dropout, applied to a multi-task chest-radiograph classifier (eight thoracic findings, 137,593 training images), provides an epistemic uncertainty signal that tracks generalisation across training-set scales and flags confident yet error-prone predictions. Adding this signal to the point prediction raised error-detection AUROC from 0.74 to 0.77 ($\Delta$AUROC +0.023, 95% CI [+0.014, +0.033]). In a controlled 2x2 factorial experiment, a clinical-decision-support agent exploited this uncertainty only when it was delivered as a binary error-risk flag rather than as raw scores, cutting confident misdiagnoses on unreliable findings from 8.5% to 2.7%. Epistemic uncertainty estimation thus carries decision-relevant information beyond point predictions, but its value for downstream agents depends on how it is communicated.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 7 canonical work pages

  1. [1]

    Diagnostic error and bias in the department of radiology: a pictorial essay.Insights into Imaging14, 163 (2023)

    Zhang L, Wen X, Li JW, Jiang X, Yang XF, Li M. Diagnostic error and bias in the department of radiology: a pictorial essay.Insights into Imaging14, 163 (2023). https://doi.org/10.1186/s13244-023-01521-7

    doi:10.1186/s13244-023-01521-7 2023

  2. [2]

    Mod- eling clinical uncertainty in radiology reports: from explicit uncertainty markers to implicit reasoning pathways

    Rabaey P, Moon JH, Lee JO, Kim MG, Yoon H, Demeester T, Choi E. Mod- eling clinical uncertainty in radiology reports: from explicit uncertainty markers to implicit reasoning pathways. In:Proceedings of the Language Resources and Evaluation Conference (LREC), 11852–11873. European Language Resources Association (ELRA), 2026. https://doi.org/10.63317/2i8nrgcnb52p

    work page doi:10.63317/2i8nrgcnb52p 2026

  3. [3]

    Mitigating overconfidence in large language models: a behavioral lens on confidence estimation and calibration

    Wen B, Xu C, Han B, Wolfe R, Wang LL, Howe B. Mitigating overconfidence in large language models: a behavioral lens on confidence estimation and calibration. In:NeurIPS 2024 Workshop on Behavioral Machine Learning, 2024. https:// openreview.net/forum?id=y9UdO5cmHs

    2024

  4. [4]

    Quantifying uncert-AI-nty: testing the accuracy of LLMs’ confidence judgments.Memory & Cognition54, 375–400 (2025)

    Cash TN, Oppenheimer DM, Christie S, Devgan M. Quantifying uncert-AI-nty: testing the accuracy of LLMs’ confidence judgments.Memory & Cognition54, 375–400 (2025). https://doi.org/10.3758/s13421-025-01755-4

    work page doi:10.3758/s13421-025-01755-4 2025

  5. [5]

    Learn to be uncertain: leveraging uncertain labels in chest X-rays with Bayesian neural networks

    Yang HY, Yang J, Pan Y, Cao K, Song Q, Gao F, Yin Y. Learn to be uncertain: leveraging uncertain labels in chest X-rays with Bayesian neural networks. In:Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR) Workshops, 5–8, 2019. https://openaccess.thecvf.com/content CVPR W2019/html/Uncertainty and Robustness in Deep V...

    2019

  6. [6]

    Uncertainty quan- tification in segmenting tuberculosis-consistent findings in frontal chest X-rays

    Rajaraman S, Zamzmi G, Yang F, Xue Z, Jaeger S, Antani SK. Uncertainty quan- tification in segmenting tuberculosis-consistent findings in frontal chest X-rays. Biomedicines10(6), 1323 (2022). https://doi.org/10.3390/biomedicines10061323

    work page doi:10.3390/biomedicines10061323 2022

  7. [7]

    Integrating uncertainty in deep neural networks for MRI based stroke analysis.Medical Image Analysis65, 101790 (2020)

    Herzog L, Murina E, D¨ urr O, Wegener S, Sick B. Integrating uncertainty in deep neural networks for MRI based stroke analysis.Medical Image Analysis65, 101790 (2020). https://doi.org/10.1016/j.media.2020.101790

    arXiv 2020

  8. [8]

    Improving uncertainty estimation with semi-supervised deep learning for COVID-19 detection using chest X-ray images.IEEE Access9, 85442–85454 (2021)

    Calderon-Ramirez S, Yang S, Moemeni A, Colreavy-Donnelly S, Elizondo DA, Oala L, Rodr ´ ıguez-Capit´ an J, Jim´ enez-Navarro M, L´ opez-Rubio E, Molina- Cabello MA. Improving uncertainty estimation with semi-supervised deep learning for COVID-19 detection using chest X-ray images.IEEE Access9, 85442–85454 (2021). https://doi.org/10.1109/ACCESS.2021.3085418

    arXiv 2021

Show all 27 references
  1. [9]

    Bargagna F, De Santi LA, Martini N, Genovesi D, Favilli B, Vergaro G, Emdin M, Giorgetti A, Positano V, Santarelli MF. Bayesian convolutional neural networks in medical imaging classification: a promising solution for deep learning limits in data scarcity scenarios.Journal of ...

    2023 doi

  2. [10]

    A systematic comparison of Bayesian deep learning robustness in diabetic retinopathy tasks

    Filos A, Farquhar S, Gomez AN, Rudner TGJ, Kenton Z, Smith L, Alizadeh M, de Kroon A, Gal Y. A systematic comparison of Bayesian deep learning robustness in diabetic retinopathy tasks. Preprint at https://arxiv.org/abs/1912.10481 (2019)

    1912 arXiv

  3. [11]

    Bayesian uncertainty estimation for detection of long-tailed and unseen conditions in medical images.Journal of Medical Imaging10(5), 054501 (2023)

    Rezaei M, N¨ appi JJ, Bischl B, Yoshida H. Bayesian uncertainty estimation for detection of long-tailed and unseen conditions in medical images.Journal of Medical Imaging10(5), 054501 (2023). https://doi.org/10.1117/1.JMI.10.5. 054501

    2023 doi

  4. [12]

    Wimmer L, Sale Y, Hofman P, Bischl B, H¨ ullermeier E. Quantifying aleatoric and epistemic uncertainty in machine learning: are conditional entropy and mutual information appropriate measures? In:Proceedings of the 39th Conference on Uncertainty in Artificial Intelligence (UAI...

    2023

  5. [13]

    Dropout as a Bayesian approximation: representing model uncertainty in deep learning

    Gal Y, Ghahramani Z. Dropout as a Bayesian approximation: representing model uncertainty in deep learning. In:Proceedings of the 33rd International Conference on Machine Learning (ICML), PMLR48, 1050–1059, 2016. https://proceedings. mlr.press/v48/gal16.html

    2016

  6. [14]

    A survey of uncertainty in deep neural networks.Artificial Intelligence Review56, 1513–1589 (2023)

    Gawlikowski J, Tassi CRN, Ali M, Lee J, Humt M, Feng J, Kruspe A, Triebel R, Jung P, Roscher R, Shahzad M, Yang W, Bamler R, Zhu XX. A survey of uncertainty in deep neural networks.Artificial Intelligence Review56, 1513–1589 (2023). https://doi.org/10.1007/s10462-023-10562-9

    2023 doi

  7. [15]

    Simple and scalable predictive uncer- tainty estimation using deep ensembles

    Lakshminarayanan B, Pritzel A, Blundell C. Simple and scalable predictive uncer- tainty estimation using deep ensembles. In:Advances in Neural Information Processing Systems (NeurIPS)30, 6405–6416, 2017. https://proceedings.neurips. cc/paper/2017/hash/9ef2ed4b7fd2c810847ffa5fa...

    2017

  8. [16]

    Evaluating scalable Bayesian deep learning methods for robust computer vision

    Gustafsson FK, Danelljan M, Sch¨ on TB. Evaluating scalable Bayesian deep learning methods for robust computer vision. In:Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 318–319, 2020. https://doi.org/10.1109/CVPR W50498.2020.00167

    2020

  9. [17]

    How certain is your Transformer? In:Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics (EACL), 1833–1840, 2021

    Shelmanov A, Tsymbalov E, Puzyrev D, Fedyanin K, Panchenko A, Panov M. How certain is your Transformer? In:Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics (EACL), 1833–1840, 2021. https://doi.org/10.18653/v1/2021.eacl...

    2021 doi

  10. [18]

    Foundation models for generalist medical artificial intelligence.Nature616, 259–265 (2023)

    Moor M, Banerjee O, Abad ZSH, Krumholz HM, Leskovec J, Topol EJ, Rajpurkar P. Foundation models for generalist medical artificial intelligence.Nature616, 259–265 (2023). https://doi.org/10.1038/s41586-023-05881-4 13

    2023 doi

  11. [19]

    Towards generalist biomedical AI.NEJM AI1(3), AIoa2300138 (2024)

    Tu T, Azizi S, Driess D, Schaekermann M, Amin M, Chang PC, Carroll A, Lau C, Tanno R, Ktena I, Palepu A, Mustafa B, Chowdhery A, Liu Y, Kornblith S, Fleet D, Mansfield P, Prakash S, Wong R, Virmani S, Semturs C, Mahdavi SS, Green B, Dominowska E, Aguera y Arcas B, Barral J, We...

    2024 doi

  12. [20]

    Development and validation of an autonomous artificial intelli- gence agent for clinical decision-making in oncology.Nature Cancer6, 1337–1349 (2025)

    Ferber D, El Nahhas OSM, W¨ olflein G, Wiest IC, Clusmann J, Leßmann ME, Foersch S, Lammert J, Tschochohei M, J¨ ager D, Salto-Tellez M, Schultz N, Truhn D, Kather JN. Development and validation of an autonomous artificial intelli- gence agent for clinical decision-making in o...

    2025 doi

  13. [21]

    A survey of LLM-based agents in medicine: how far are we from Baymax? In:Findings of the Association for Computational Linguistics: ACL 2025, 10345–10359, 2025

    Wang W, Ma Z, Wang Z, Wu C, Ji J, Chen W, Li X, Yuan Y. A survey of LLM-based agents in medicine: how far are we from Baymax? In:Findings of the Association for Computational Linguistics: ACL 2025, 10345–10359, 2025. https://doi.org/10.18653/v1/2025.findings-acl.539

    2025 doi

  14. [22]

    DINOv2: learning robust visual features without supervision.Transactions on Machine Learning Research(2024)

    Oquab M, Darcet T, Moutakanni T, Vo H, Szafraniec M, Khalidov V, Fernandez P, Haziza D, Massa F, El-Nouby A, Assran M, Ballas N, Galuba W, Howes R, Huang PY, Li SW, Misra I, Rabbat M, Sharma V, Synnaeve G, Xu H, Jegou H, Mairal J, Labatut P, Joulin A, Bojanowski P. DINOv2: lea...

    2024

  15. [23]

    A comprehensive bedside chest radiography dataset with structured, itemized and graded radiologic reports

    Truhn, D., Geiger, D., Siepmann, R., von der St¨ uck, M.S., Bressem, K.K., Kather, J.N., Kuhl, C., M¨ uller-Franzes, G., & Nebelung, S. A comprehensive bedside chest radiography dataset with structured, itemized and graded radiologic reports. Scientific Data13, 632 (2026). htt...

    2026 doi

  16. [24]

    Auto-encoding variational Bayes

    Kingma DP, Welling M. Auto-encoding variational Bayes. In:Proceedings of the 2nd International Conference on Learning Representations (ICLR), 2014. Preprint at https://arxiv.org/abs/1312.6114 (2013)

    2014 arXiv

  17. [25]

    Deep learning improves physician accuracy in the comprehensive detection of abnormalities on chest X-rays.Scientific Reports14, 25151 (2024)

    Anderson PG, Tarder-Stoll H, Alpaslan M, Keathley N, Levin DL, Venkatesh S, Bartel E, Sicular S, Howell S, Lindsey R V, Jones RM. Deep learning improves physician accuracy in the comprehensive detection of abnormalities on chest X-rays.Scientific Reports14, 25151 (2024). https...

    2024

  18. [26]

    Error and discrepancy in radiology: inevitable or avoidable?Insights into Imaging8, 171–182 (2017)

    Brady AP. Error and discrepancy in radiology: inevitable or avoidable?Insights into Imaging8, 171–182 (2017). https://doi.org/10.1007/s13244-016-0534-1

    2017 doi

  19. [27]

    Uncertainty estimation in medical image classification: systematic review.JMIR Medical Informatics10(8), e36427 (2022)

    Kurz A, Hauser K, Mehrtens HA, Krieghoff-Henning E, Hekler A, Kather JN, Fr¨ ohling S, von Kalle C, Brinker TJ. Uncertainty estimation in medical image classification: systematic review.JMIR Medical Informatics10(8), e36427 (2022). 14 https://doi.org/10.2196/36427 15

    2022 doi

Pith tools

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