Pith. sign in

REVIEW 5 major objections 5 minor 55 references

Evaluating and Mitigating Bias in AI-Based Medical Text Generation

T0 review · 5 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Medical text generators show measurable bias across sex, race, and age; a selective training method cuts those gaps by over 30% without dropping quality.

desk verdict Useful first systematic evaluation of bias in medical text generation, but the headline 30% reduction rests on an underspecified pair-based metric that needs repair. read the letter →

arxiv 2504.17279 v1 pith:ISEXS4SO submitted 2025-04-24 cs.CL

Xiuying Chen , Tairan Wang , Juexiao Zhou , Zirui Song , Xin Gao , Xiangliang Zhang This is my paper · ORCID
classification cs.CL
keywords fairnessmedicaltextgenerationradiologyreportsummarizationlargelanguagemodelsbiasmitigationselectiveoptimizationmetric-awaredifference
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 asks whether AI systems that write medical text—radiology impression reports, report summaries, and scientific-paper abstracts—serve all demographic groups equally. Across three tasks, three model scales, and six evaluation metrics, it finds significant and often intersectional performance gaps by sex, race, and age. To close these gaps, the paper proposes a training-time selection algorithm that focuses backpropagation on the cases where the model is most wrong, judged both by word-level cross-entropy loss and by a ranking loss over pathology-aware candidate quality. The central claim is that this selective optimization reduces the average fairness gap by more than 30% while keeping overall generation accuracy within about 2% of baseline. A sympathetic reader should care because the bias is not hypothetical: the paper shows concrete missed diagnoses, such as an omitted cardiomegaly finding in a male patient's generated impression.

What carries the argument

Two mechanisms carry the argument. The first is the Metric-aware Fairness Difference (MFD), a per-metric gap score computed as the average absolute difference between paired per-instance metric values of two subgroups, which turns fairness into an explicit training signal and evaluation statistic. The second is selective optimization: the model computes, for each case in a batch, a cross-entropy generation loss plus a ranking loss that penalizes the model for scoring low-quality candidate references above high-quality ones, then backpropagates only on the top fraction of cases by that combined loss. The ranking loss is what makes the selection medical-aware, because candidate quality is defined by both ROUGE word overlap and CheXpert pathology-observation accuracy, and the whole selection procedure remains differentiable because it is just weighting the loss of selected cases.

What would settle it

Recompute the reported average MFD for both baseline and proposed models after replacing the paper's implicit pairing with a permutation-invariant comparison (for example, the Wasserstein distance between the two subgroups' per-instance metric distributions), using the released code and test splits; if the proposed method no longer shows a 30% or larger reduction, the mitigation claim rests on the unspecified pairing rather than on genuine fairness improvement.

Watch

Extended reading notes

Core claim

The paper's central claim is that unfairness in medical text generation is widespread, measurable, and reducible without sacrificing quality. It introduces the Metric-aware Fairness Difference (MFD), which averages the absolute difference between per-instance scores of two subgroups under a given metric, and uses it to show that existing models produce significant gaps on ROUGE and CheXpert scores across sex, race, age, and species, with larger gaps for intersectional groups such as Black males. Its proposed remedy is a selection-based optimization loop: within each batch, only the hardest cases—those with the highest combined cross-entropy loss and ranking loss over candidate reference quality—are used for backpropagation. Across R2Gen, BART-base, BART-large, and LoRA-tuned LLaMA2-13B, and across radiology report generation, report summarization, and paper summarization, the method reports an average 35.27% reduction in MFD and a relative accuracy change typically within 2%, with no significant performance drop in most settings.

Load-bearing premise

The quantitative claim that unfairness shrinks by more than 30% depends on MFD, whose per-instance pairing between subgroups of different sizes is never defined; if the pairing scheme is arbitrary, the reported reduction could partly be an artifact of how the metric is computed.

Editorial extensions

If this is right

  • Training medical text generators with loss-based hard-case selection can reduce demographic performance gaps by more than 30% on average while holding overall metric changes within about 2%.
  • The same mechanism reduces compounded disparities for intersectional groups such as Black males, not just for single attributes.
  • The approach transfers from task-specific models with around 100M parameters to a 13B-parameter LLM fine-tuned with LoRA, so it is not restricted to small architectures.
  • Because the selection criteria combine word-level and pathology-level losses, the method can be adapted to fairness in any evaluation metric by swapping the metric that defines candidate quality.
  • In some settings the fairness-focused training also narrows confidence intervals and improves scores for underperforming groups, suggesting fairness and robustness can move together.

Reading between the lines

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

  • Because selection is driven by high combined loss rather than by demographic identity, the mechanism may work by reweighting hard examples generally; the fairness gains could be a side effect of focusing training on cases where all groups are weak, with underrepresented groups happening to contain more such cases.
  • The paper's oversampling study suggests that simply balancing group sizes does not fix the gap, so the mechanism is not equivalent to data rebalancing; this points to difficulty-based selection rather than representation-based selection as the operative cause.
  • If MFD is the target, a natural extension is to make the pairing explicit and distribution-free, or to train directly against a permutation-invariant fairness penalty, which would make the reported reductions robust to implementation details.
  • The same selection framework could be transferred to other high-stakes generation domains, such as legal document drafting or clinical trial protocols, where word-level accuracy alone misses the domain-specific concepts that matter.
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

5 major / 5 minor

Summary. The paper investigates fairness in medical text generation, evaluating bias across sex, race, age, and intersectional groups on radiology report generation, report summarization, and paper summarization tasks. It proposes a selection-based optimization algorithm that trains only on high-loss cases, where the loss combines cross-entropy with a ranking loss that scores pre-generated candidate outputs by ROUGE and CheXpert. The authors report that the proposed method reduces the Metric-aware Fairness Difference (MFD) by over 30% on average while keeping overall generation quality within 2% of the original model. They also claim that the approach is model- and task-agnostic, demonstrating it on R2Gen, BART-base/large, and LLaMA2-13B with LoRA.

Significance. If the central claims hold, this paper makes a useful contribution by being among the first to systematically evaluate and mitigate bias in medical text generation, covering multiple tasks, model scales, and demographic axes. The release of code and the use of several public datasets support reproducibility. The proposed selective optimization is simple and could be adopted broadly. However, the significance is conditional on resolving the ambiguity in the definition of MFD and on demonstrating that the observed fairness improvements are not artifacts of the measurement or of metric-specific overfitting. The paper also provides a thorough evaluation of bias across intersectional groups, which is valuable for the fairness community.

major comments (5)
  1. [Sec. 5.1, Eq. (9)] The MFD definition in Eq. (9) is an average absolute difference between paired instances of two subgroups, but the paper never specifies how pairs are formed when subgroup sizes differ. The datasets used have unequal subgroup sizes (e.g., 50,273 males vs. 51,965 females in MIMIC-CXR; age groups of 6,749 vs. 2,077 in PubMed). If n is the smaller size and pairing is done by dataset order, the metric is sensitive to an arbitrary ordering; if n is intended to be the full group size, the sum is not well-defined. Consequently, the headline quantitative claim of a 35.27% average MFD reduction and the abstract's "more than 30%" statement are not reproducible as written. The authors must specify the pairing rule (e.g., random pairing, rank-based alignment, or a distributional alternative such as the absolute difference of group means) and justify that the chosen definition measures meaningful group disparity.
  2. [Sec. 5.1 and Fig. 6] The Mann-Whitney U test is used to assess whether MFD is significantly reduced, but MFD as defined in Eq. (9) is a single scalar per metric and group pair, not a sample from a distribution. To run the reported test, the authors must be applying the test to per-pair absolute differences (the summands in Eq. (9)), which again requires the pairing to be explicitly specified. As written, the significance claims in Fig. 6 are unsupported because the underlying statistic is ambiguous. Please clarify the exact quantity being tested and report the associated sample size.
  3. [Sec. 4.1, Eqs. (1)-(3)] The proposed method depends on several hyperparameters that are never reported: the selection ratio gamma, the ranking margins delta_i, and the number of candidates n. Section 5.2 gives optimization details but omits these values for every dataset and model. Without them, the algorithm cannot be re-implemented or compared against. Please report the hyperparameter values used, and ideally include a sensitivity analysis for gamma, which directly controls how much training signal is discarded.
  4. [Sec. 4.1 and Sec. 2.1/5.1] The ranking loss in Eq. (2) uses ROUGE and CheXpert scores of pre-generated candidates as quality labels, and the same metrics are used to compute MFD, the fairness measure. This means the training objective directly optimizes the metric family used for evaluation. While not circular in the sense of using group labels, it raises the risk that the reported fairness improvements are metric-specific overfitting rather than a general reduction in bias. The paper would be strengthened by evaluating on a held-out metric not used in training (e.g., BERTScore, which is mentioned in Section 2.1) or by reporting an ablation that removes the ranking-loss component to show that the improvement does not depend solely on optimizing the evaluation metrics.
  5. [Sec. 4.2, Figs. 6-7] The only baseline compared is the vanilla model without any fairness intervention. To support the claim that the proposed selective optimization is an effective mitigation algorithm, the authors should compare against at least one standard fairness approach, such as group-based reweighting, group-DRO, or a straightforward hard-example mining baseline that uses only the cross-entropy loss. This comparison is necessary to determine whether the observed reductions in MFD are attributable to the specific selection rule or merely to any form of selective training on difficult examples.
minor comments (5)
  1. [Sec. 2.3] The phrase "We filtered out unpaired cases" is unclear; please specify what is unpaired (e.g., images without reports, or reports without corresponding images) and how many cases were removed.
  2. [Fig. 3 caption and Sec. 5.1] The caption reads "two-side Mann-Whitney U test"; this should be "two-sided".
  3. [Sec. 4.1, Fig. 5] The notation "P1>...>PN?" in Fig. 5 is not defined; it should be clear that score(Ci) is the model's generation probability (or log-likelihood) for candidate i.
  4. [Sec. 5.2] For the LLaMA2-13B LoRA experiments, please report the LoRA rank, alpha, and dropout values, as these affect reproducibility.
  5. [Abstract and Sec. 2.1] The paper mentions "six generation evaluation metrics" in the introduction; please explicitly list them (likely ROUGE-1, ROUGE-2, ROUGE-L, and CheXpert precision, recall, F1) in Section 2.1.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the reported fairness gains are empirical and not forced by construction, though the training objective shares metric families with the evaluation and a few minor self-citations are present.

full rationale

The paper's central claims are empirical evaluations rather than first-principles derivations. MFD (Eq. 9) is introduced as a measurement definition, not as the output of a derivation; the selection algorithm (Eqs. 1-3) does not use group labels or MFD, so its reported fairness gains are not forced by construction. The ranking loss (Eq. 2) does use ROUGE and CheXpert scores of pre-generated candidates as quality labels, and the same metric families are used for evaluation, which creates a risk of overfitting to the metrics rather than a circular derivation; the group-level MFD reductions are still emergent since selection is loss-based and group-agnostic. Minor self-citations ([37], [40]) support routine claims about ROUGE and are not load-bearing. The unspecified pairing in Eq. 9 for unequal subgroup sizes is a reproducibility or correctness concern, not a circularity, so it does not elevate the circularity score.

Assumptions & free parameters 3 free parameters · 3 assumptions · 1 invented entities

The central claim (over 30% MFD reduction, under 2% accuracy change) depends on three unreported free parameters (selection ratio, ranking margins, candidate count), on the validity of ROUGE/CheXpert as quality measures, and on the newly introduced MFD metric, whose construction is underspecified. The method itself does not use demographic labels, so there are no fitted group-specific parameters.

free parameters (3)
  • selection ratio gamma = not reported
    Controls the fraction of highest-loss cases selected for backpropagation in Eq. 1 and Eq. 3. No value or sensitivity analysis is given in the main text, and the reported fairness reductions likely depend on this choice.
  • ranking loss margin delta_i = not reported
    Margin between candidate scores in Eq. 2. Chosen by hand, but its value and effect on results are not disclosed.
  • number of candidates n = not reported
    Number of pre-generated candidate texts used in the ranking loss. Not specified in the methods section.
assumptions (3)
  • domain assumption ROUGE and CheXpert scores are valid proxies for the quality and clinical accuracy of generated medical text.
    The entire evaluation, including the MFD fairness metric, is built on these metrics. The paper cites prior validation for CheXpert, but does not independently validate that optimizing these metrics improves patient-relevant outcomes.
  • ad hoc to paper Selecting cases with high cross-entropy and ranking loss preferentially selects underperforming demographic groups, and training on them reduces group disparities.
    This is the core mechanism of the proposed mitigation, stated in Section 4. It is an empirical heuristic; the paper provides no theoretical argument or explicit demonstration that high-loss cases are disproportionately from disadvantaged groups.
  • standard math The Mann-Whitney U test assumptions (independent samples, ordinal metric values) hold for the subgroup comparisons.
    Used for all significance tests in Figs. 3, 4, 6, 7, 8. Standard for comparing two groups on a continuous metric, and the samples are disjoint patients or papers.
invented entities (1)
  • Metric-aware Fairness Difference (MFD)
    purpose: Quantify unfairness in text generation as the average absolute difference between per-instance metric values of two subgroups.
    A new evaluation quantity introduced by this paper. It is not benchmarked against human judgments or external fairness metrics, and its pairing procedure for unequal group sizes is unspecified, so its validity as a fairness measure is not independently established.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Evaluating and Mitigating Bias in AI-Based Medical Text Generation." pith.science (2026). https://pith.science/paper/ISEXS4SO

@misc{pith2026250417279,
  author       = {Pith},
  title        = {Pith review of: Evaluating and Mitigating Bias in AI-Based Medical Text Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ISEXS4SO}},
  note         = {Machine review of arXiv:2504.17279}
}
read the original abstract

Artificial intelligence (AI) systems, particularly those based on deep learning models, have increasingly achieved expert-level performance in medical applications. However, there is growing concern that such AI systems may reflect and amplify human bias, and reduce the quality of their performance in historically under-served populations. The fairness issue has attracted considerable research interest in the medical imaging classification field, yet it remains understudied in the text generation domain. In this study, we investigate the fairness problem in text generation within the medical field and observe significant performance discrepancies across different races, sexes, and age groups, including intersectional groups, various model scales, and different evaluation metrics. To mitigate this fairness issue, we propose an algorithm that selectively optimizes those underperformed groups to reduce bias. The selection rules take into account not only word-level accuracy but also the pathology accuracy to the target reference, while ensuring that the entire process remains fully differentiable for effective model training. Our evaluations across multiple backbones, datasets, and modalities demonstrate that our proposed algorithm enhances fairness in text generation without compromising overall performance. Specifically, the disparities among various groups across different metrics were diminished by more than 30% with our algorithm, while the relative change in text generation accuracy was typically within 2%. By reducing the bias generated by deep learning models, our proposed approach can potentially alleviate concerns about the fairness and reliability of text generation diagnosis in medical domain. Our code is publicly available to facilitate further research at https://github.com/iriscxy/GenFair.

Figures

Figures reproduced from arXiv: 2504.17279 by the authors.

Figure 1
Figure 1. An example where both male and female have similar input, yet the text generation model neglects cardiomegaly in the male case. In this study, we first evaluate the presence of unfairness issues in image-based computer-aided diagnosis, text-based radiology report, and medical paper summarization using publicly available datasets (Tab. 1). Our evaluation spans six generation evaluation metrics and three different sca… view at source ↗
Figure 2
Figure 2. Model Pipeline: Our approach involves three tasks: radiology report generation, report summarization, and paper summarization, using the public benchmark MIMIC-CXR and PubMed datasets. We trained a deep-learning text generation model using our selective optimization mechanism, which considers two key aspects: word-level overlap and pathology observation. We evaluated pairwise fairness across different subgroups, inc… view at source ↗
Figure 3
Figure 3. Performance disparities were observed across various categories, including sex (female vs. male), race (black vs. white), age (young vs. aged), and species (humans vs. animals), for ROUGE comparison in: (a) radiology report generation, (b) summarization on the CXR dataset, and (c) summarization on the PubMed dataset; and for CheXpert comparison in: (d) radiology report generation and (e) report summarization. Signif… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Performance disparities were observed across intersectional groups in: (a) radiology report generation, and (b)(c) report summa￾rization. The top row represents baseline performance, while the bottom row illustrates the enhanced performance achieved by incorporating ou…
Figure 5
Figure 5. Figure 5: The details of the selection algorithm. The encoder-decoder or decoder-only model outputs the generation probability for the reference as well as a set of candidates. The ranking loss evaluates if the model assigns a greater generation probability to candidates of high…
Figure 6
Figure 6. Figure 6: Our proposed method significantly reduced the Metric-aware Fairness Difference (MFD) across various pairwise comparisons, such as female vs. male, as shown (a) radiology report generation, (b) report summarization, and (c) scholar paper summarization. Significant diffe…
Figure 7
Figure 7. Figure 7: Our proposed method maintains overall performance across different tasks: (a) radiology report generation, (b) report summariza￾tion, and (c) scholar paper summarization. Significant differences denoted by asterisks (*), were determined using the two-side Mann-Whitney …
Figure 8
Figure 8. Figure 8: (a) Significant performance differences were observed for the LLM Llama2-13B. (b) Our method significantly reduced the MFD for Llama2. (c) Our method does not adversely affect the overall performance of the Llama2. is determined by comparing the metric values of intere…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 42 canonical work pages

  1. [1]

    Artificial intelligence in healthcare: past, present and future,

    F. Jiang, Y. Jiang, H. Zhi, Y. Dong, H. Li, S. Ma, Y. Wang, Q. Dong, H. Shen, and Y. Wang, “Artificial intelligence in healthcare: past, present and future,” Stroke and vascular neurology , vol. 2, no. 4, 2017

    work page 2017

  2. [2]

    Chexnet: Radiologist- level pneumonia detection on chest x-rays with deep learning,

    P . Rajpurkar, J. Irvin, K. Zhu, B. Yang, H. Mehta, T. Duan, D. Ding, A. Bagul, C. Langlotz, K. Shpanskaya et al., “Chexnet: Radiologist- level pneumonia detection on chest x-rays with deep learning,” arXiv preprint arXiv:1711.05225, 2017

    arXiv 2017

  3. [3]

    Automated diagnosing primary open- angle glaucoma from fundus image by simulating human’s grad- ing with deep learning,

    M. Lin, B. Hou, L. Liu, M. Gordon, M. Kass, F. Wang, S. H. Van Tassel, and Y. Peng, “Automated diagnosing primary open- angle glaucoma from fundus image by simulating human’s grad- ing with deep learning,” Scientific reports, vol. 12, no. 1, p. 14080, 2022

    work page 2022

  4. [4]

    Radiologist shortage leaves patient care at risk, warns royal college,

    A. Rimmer, “Radiologist shortage leaves patient care at risk, warns royal college,” BMJ: British Medical Journal (Online), vol. 359, 2017

    work page 2017

  5. [5]

    Scholarchemqa: Unveiling the power of language models in chemical research question answering,

    X. Chen, T. Wang, T. Guo, K. Guo, J. Zhou, H. Li, M. Zhuge, J. Schmidhuber, X. Gao, and X. Zhang, “Scholarchemqa: Unveiling the power of language models in chemical research question answering,” arXiv preprint arXiv:2407.16931, 2024

    arXiv 2024

  6. [6]

    Nature of metal-support interaction for metal catalysts on oxide supports,

    T. Wang, J. Hu, R. Ouyang, Y. Wang, Y. Huang, S. Hu, and W.-X. Li, “Nature of metal-support interaction for metal catalysts on oxide supports,” Science, vol. 386, no. 6724, pp. 915–920, 2024

    work page 2024

  7. [7]

    Dis- secting racial bias in an algorithm used to manage the health of populations,

    Z. Obermeyer, B. Powers, C. Vogeli, and S. Mullainathan, “Dis- secting racial bias in an algorithm used to manage the health of populations,” Science, vol. 366, no. 6464, pp. 447–453, 2019

    work page 2019

  8. [8]

    Chexclusion: Fairness gaps in deep chest x-ray classifiers,

    L. Seyyed-Kalantari, G. Liu, M. McDermott, I. Y. Chen, and M. Ghassemi, “Chexclusion: Fairness gaps in deep chest x-ray classifiers,” in BIOCOMPUTING 2021: proceedings of the Pacific symposium. World Scientific, 2020, pp. 232–243

    work page 2021

Show all 55 references
  1. [9]

    Medfair: Benchmarking fair- ness for medical imaging,

    Y. Zong, Y. Yang, and T. Hospedales, “Medfair: Benchmarking fair- ness for medical imaging,” in The Eleventh International Conference on Learning Representations, 2022

    2022

  2. [10]

    Improving model fairness in image-based computer-aided diagnosis,

    M. Lin, T. Li, Y. Yang, G. Holste, Y. Ding, S. H. Van Tassel, K. Ko- vacs, G. Shih, Z. Wang, Z. Lu et al., “Improving model fairness in image-based computer-aided diagnosis,” Nature Communications, vol. 14, no. 1, p. 6261, 2023

    2023

  3. [11]

    Gender imbalance in medical imaging datasets produces biased classifiers for computer-aided diagnosis,

    A. J. Larrazabal, N. Nieto, V . Peterson, D. H. Milone, and E. Fer- rante, “Gender imbalance in medical imaging datasets produces biased classifiers for computer-aided diagnosis,” Proceedings of the National Academy of Sciences , vol. 117, no. 23, pp. 12 592–12 594, 2020

    2020

  4. [12]

    Rad- fusion: Benchmarking performance and fairness for multimodal pulmonary embolism detection from ct and ehr,

    Y. Zhou, S.-C. Huang, J. A. Fries, A. Youssef, T. J. Amrhein, M. Chang, I. Banerjee, D. Rubin, L. Xing, N. Shah et al. , “Rad- fusion: Benchmarking performance and fairness for multimodal pulmonary embolism detection from ct and ehr,” arXiv preprint arXiv:2111.11665, 2021

    2021 arXiv

  5. [13]

    Fairness of classifiers across skin tones in dermatology,

    N. M. Kinyanjui, T. Odonga, C. Cintas, N. C. Codella, R. Panda, P . Sattigeri, and K. R. Varshney, “Fairness of classifiers across skin tones in dermatology,” in International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, 2020, pp. 320–329

    2020

  6. [14]

    Evaluate underdiagnosis and overdiagnosis bias of deep learning model on primary open-angle glaucoma diagnosis in under-served populations,

    M. Lin, Y. Xiao, B. Hou, T. Wanyan, M. M. Sharma, Z. Wang, F. Wang, S. Van Tassel, and Y. Peng, “Evaluate underdiagnosis and overdiagnosis bias of deep learning model on primary open-angle glaucoma diagnosis in under-served populations,” AMIA Summits on Translational Science P...

    2023

  7. [15]

    Ethical analysis to im- prove decision-making on health technologies,

    S. I. Saarni, B. Hofmann, K. Lampe, D. L ¨uhmann, M. M ¨akel¨a, M. Velasco-Garrido, and I. Autti-R ¨am¨o, “Ethical analysis to im- prove decision-making on health technologies,” Bulletin of the World Health Organization, vol. 86, pp. 617–623, 2008

    2008

  8. [16]

    On the ethics of algorithmic decision- making in healthcare,

    T. Grote and P . Berens, “On the ethics of algorithmic decision- making in healthcare,” Journal of medical ethics , vol. 46, no. 3, pp. 205–211, 2020

    2020

  9. [17]

    Improving the fairness of chest x-ray classifiers,

    H. Zhang, N. Dullerud, K. Roth, L. Oakden-Rayner, S. Pfohl, and M. Ghassemi, “Improving the fairness of chest x-ray classifiers,” in Conference on health, inference, and learning . PMLR, 2022, pp. 204–233

    2022

  10. [18]

    Fairness without demographics through adversarially 11 reweighted learning,

    P . Lahoti, A. Beutel, J. Chen, K. Lee, F. Prost, N. Thain, X. Wang, and E. Chi, “Fairness without demographics through adversarially 11 reweighted learning,” Advances in neural information processing systems, vol. 33, pp. 728–740, 2020

    2020

  11. [19]

    Pairwise fairness for ranking and regression,

    H. Narasimhan, A. Cotter, M. Gupta, and S. Wang, “Pairwise fairness for ranking and regression,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 34, no. 04, 2020, pp. 5248– 5255

    2020

  12. [20]

    The limits of fair medical imaging ai in real-world generaliza- tion,

    Y. Yang, H. Zhang, J. W. Gichoya, D. Katabi, and M. Ghassemi, “The limits of fair medical imaging ai in real-world generaliza- tion,” Nature Medicine, pp. 1–11, 2024

    2024

  13. [21]

    An extensive benchmark study on biomed- ical text generation and mining with chatgpt,

    Q. Chen, H. Sun, H. Liu, Y. Jiang, T. Ran, X. Jin, X. Xiao, Z. Lin, H. Chen, and Z. Niu, “An extensive benchmark study on biomed- ical text generation and mining with chatgpt,” Bioinformatics, vol. 39, no. 9, p. btad557, 2023

    2023

  14. [22]

    Chatgpt in healthcare: a taxonomy and systematic review,

    J. Li, A. Dada, B. Puladi, J. Kleesiek, and J. Egger, “Chatgpt in healthcare: a taxonomy and systematic review,” Computer Methods and Programs in Biomedicine, p. 108013, 2024

    2024

  15. [23]

    Opportunities and chal- lenges for chatgpt and large language models in biomedicine and health,

    S. Tian, Q. Jin, L. Yeganova, P .-T. Lai, Q. Zhu, X. Chen, Y. Yang, Q. Chen, W. Kim, D. C. Comeau et al., “Opportunities and chal- lenges for chatgpt and large language models in biomedicine and health,” Briefings in Bioinformatics, vol. 25, no. 1, p. bbad493, 2024

    2024

  16. [24]

    Interactive and explainable region-guided radiology report generation,

    T. Tanida, P . M ¨uller, G. Kaissis, and D. Rueckert, “Interactive and explainable region-guided radiology report generation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 7433–7442

    2023

  17. [25]

    Radadapt: Radiology report summarization via lightweight domain adaptation of large language models,

    D. Van Veen, C. Van Uden, M. Attias, A. Pareek, C. Bluethgen, M. Polacin, W. Chiu, J.-B. Delbrouck, J. Z. Chaves, C. Langlotz et al., “Radadapt: Radiology report summarization via lightweight domain adaptation of large language models,” in The 22nd Work- shop on Biomedical Nat...

    2023

  18. [26]

    The advent of generative language models in medical education,

    M. Karabacak, B. B. Ozkara, K. Margetis, M. Wintermark, and S. Bisdas, “The advent of generative language models in medical education,” JMIR Medical Education, vol. 9, p. e48163, 2023

    2023

  19. [27]

    The next generation of evidence-based medicine,

    V . Subbiah, “The next generation of evidence-based medicine,” Nature medicine, vol. 29, no. 1, pp. 49–58, 2023

    2023

  20. [28]

    Ethical and social risks of harm from language models,

    L. Weidinger, J. Mellor, M. Rauh, C. Griffin, J. Uesato, P .-S. Huang, M. Cheng, M. Glaese, B. Balle, A. Kasirzadeh et al. , “Ethical and social risks of harm from language models,” arXiv preprint arXiv:2112.04359, 2021

    2021 arXiv

  21. [29]

    Smartphone-based conversational agents and responses to questions about mental health, interpersonal violence, and physical health,

    A. S. Miner, A. Milstein, S. Schueller, R. Hegde, C. Mangurian, and E. Linos, “Smartphone-based conversational agents and responses to questions about mental health, interpersonal violence, and physical health,” JAMA internal medicine , vol. 176, no. 5, pp. 619– 625, 2016

    2016

  22. [30]

    Patient and consumer safety risks when using conversational assistants for medical information: an observational study of siri, alexa, and google assistant,

    T. W. Bickmore, H. Trinh, S. Olafsson, T. K. O’Leary, R. Asadi, N. M. Rickles, and R. Cruz, “Patient and consumer safety risks when using conversational assistants for medical information: an observational study of siri, alexa, and google assistant,” Journal of medical Interne...

    2018

  23. [31]

    Auto- mated radiology report generation: A review of recent advances,

    P . Sloan, P . Clatworthy, E. Simpson, and M. Mirmehdi, “Auto- mated radiology report generation: A review of recent advances,” IEEE Reviews in Biomedical Engineering, 2024

    2024

  24. [32]

    A survey on automatic generation of medical imaging reports based on deep learning,

    T. Pang, P . Li, and L. Zhao, “A survey on automatic generation of medical imaging reports based on deep learning,” BioMedical Engineering OnLine, vol. 22, no. 1, p. 48, 2023

    2023

  25. [33]

    Bertscore: Evaluating text generation with bert,

    T. Zhang, V . Kishore, F. Wu, K. Q. Weinberger, and Y. Artzi, “Bertscore: Evaluating text generation with bert,” in Proc. of ICLR, 2020

    2020

  26. [34]

    Evaluation of text genera- tion: A survey,

    A. Celikyilmaz, E. Clark, and J. Gao, “Evaluation of text genera- tion: A survey,” arXiv preprint arXiv:2006.14799, 2020

    2006 arXiv

  27. [35]

    Gptscore: Evaluate as you desire,

    J. Fu, S.-K. Ng, Z. Jiang, and P . Liu, “Gptscore: Evaluate as you desire,” arXiv preprint arXiv:2302.04166, 2023

    2023 arXiv

  28. [36]

    Rouge: A package for automatic evaluation of sum- maries,

    C.-Y. Lin, “Rouge: A package for automatic evaluation of sum- maries,” in Text summarization branches out, 2004

    2004

  29. [37]

    Flexible and adaptable summarization via expertise separation,

    X. Chen, M. Li, S. Gao, X. Cheng, Q. Zhu, R. Yan, X. Gao, and X. Zhang, “Flexible and adaptable summarization via expertise separation,” in Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , 2024, pp. 2018–2027

    2024

  30. [38]

    Revisiting the gold standard: Grounding summarization evaluation with robust human evaluation,

    Y. Liu, A. R. Fabbri, P . Liu, Y. Zhao, L. Nan, R. Han, S. Han, S. R. Joty, C.-S. Wu, C. Xiong, and D. R. Radev, “Revisiting the gold standard: Grounding summarization evaluation with robust human evaluation,” Proc. of ACL, 2023

    2023

  31. [39]

    Questeval: Summarization asks for fact-based evaluation,

    T. Scialom, P .-A. Dray, S. Lamprier, B. Piwowarski, J. Staiano, A. Wang, and P . Gallinari, “Questeval: Summarization asks for fact-based evaluation,” in Proc. of EMNLP, 2021

    2021

  32. [40]

    Rethinking scientific summarization evaluation: Grounding explainable metrics on facet-aware benchmark,

    X. Chen, T. Wang, Q. Zhu, T. Guo, S. Gao, Z. Lu, X. Gao, and X. Zhang, “Rethinking scientific summarization evaluation: Grounding explainable metrics on facet-aware benchmark,” arXiv preprint arXiv:2402.14359, 2024

    2024 arXiv

  33. [41]

    Chexpert: A large chest radiograph dataset with uncertainty labels and ex- pert comparison,

    J. Irvin, P . Rajpurkar, M. Ko, Y. Yu, S. Ciurea-Ilcus, C. Chute, H. Marklund, B. Haghgoo, R. Ball, K. Shpanskayaet al., “Chexpert: A large chest radiograph dataset with uncertainty labels and ex- pert comparison,” in Proceedings of the AAAI conference on artificial intelligen...

    2019

  34. [42]

    Cxpmrg-bench: Pre-training and benchmarking for x-ray medical report generation on chexpert plus dataset,

    X. Wang, F. Wang, Y. Li, Q. Ma, S. Wang, B. Jiang, C. Li, and J. Tang, “Cxpmrg-bench: Pre-training and benchmarking for x-ray medical report generation on chexpert plus dataset,”arXiv preprint arXiv:2410.00379, 2024

    2024 arXiv

  35. [43]

    Retrieval-based chest x-ray report generation using a pre-trained contrastive language-image model,

    M. Endo, R. Krishnan, V . Krishna, A. Y. Ng, and P . Rajpurkar, “Retrieval-based chest x-ray report generation using a pre-trained contrastive language-image model,” in Machine Learning for Health. PMLR, 2021, pp. 209–219

    2021

  36. [44]

    Baselines for chest x-ray report generation,

    W. Boag, T.-M. H. Hsu, M. McDermott, G. Berner, E. Alesentzer, and P . Szolovits, “Baselines for chest x-ray report generation,” in Machine learning for health workshop. PMLR, 2020, pp. 126–140

    2020

  37. [45]

    Generating radiology reports via memory-driven transformer,

    Z. Chen, Y. Song, T.-H. Chang, and X. Wan, “Generating radiology reports via memory-driven transformer,” in Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2020, pp. 1439–1449

    2020

  38. [46]

    Bart: Denoising sequence-to-sequence pre-training for natural language genera- tion, translation, and comprehension,

    M. Lewis, Y. Liu, N. Goyal, M. Ghazvininejad, A. Mohamed, O. Levy, V . Stoyanov, and L. Zettlemoyer, “Bart: Denoising sequence-to-sequence pre-training for natural language genera- tion, translation, and comprehension,” in Proc. of ACL, 2020

    2020

  39. [47]

    Llama 2: Open foundation and fine-tuned chat models,

    H. Touvron, L. Martin, K. Stone, P . Albert, A. Almahairi, Y. Babaei, N. Bashlykov, S. Batra, P . Bhargava, S. Bhosale et al. , “Llama 2: Open foundation and fine-tuned chat models,” arXiv preprint arXiv:2307.09288, 2023

    2023 arXiv

  40. [48]

    Mimic-cxr-jpg, a large publicly available database of labeled chest radiographs,

    A. E. Johnson, T. J. Pollard, N. R. Greenbaum, M. P . Lungren, C.-y. Deng, Y. Peng, Z. Lu, R. G. Mark, S. J. Berkowitz, and S. Horng, “Mimic-cxr-jpg, a large publicly available database of labeled chest radiographs,” arXiv preprint arXiv:1901.07042, 2019

    1901 arXiv

  41. [49]

    A discourse-aware attention model for ab- stractive summarization of long documents,

    A. Cohan, F. Dernoncourt, D. S. Kim, T. Bui, S. Kim, W. Chang, and N. Goharian, “A discourse-aware attention model for ab- stractive summarization of long documents,” in Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Lingu...

    2018

  42. [50]

    Mimic-iv,

    A. Johnson, L. Bulgarelli, T. Pollard, S. Horng, L. A. Celi, and R. Mark, “Mimic-iv,”PhysioNet. Available online at: https://physionet. org/content/mimiciv/1.0/(accessed August 23, 2021), pp. 49–55, 2020

    2021

  43. [51]

    Do hospitals provide lower quality of care to black patients for pneumonia?

    F. B. Mayr, S. Yende, G. D’Angelo, A. E. Barnato, J. A. Kellum, L. Weissfeld, D. M. Yealy, M. C. Reade, E. B. Milbrandt, and D. C. Angus, “Do hospitals provide lower quality of care to black patients for pneumonia?” Critical care medicine , vol. 38, no. 3, pp. 759–765, 2010

    2010

  44. [52]

    Transformers: 12 State-of-the-art natural language processing,

    T. Wolf, L. Debut, V . Sanh, J. Chaumond, C. Delangue, A. Moi, P . Cistac, T. Rault, R. Louf, M. Funtowicz et al. , “Transformers: 12 State-of-the-art natural language processing,” in Proc. of EMNLP , 2020

    2020

  45. [53]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

    2016

  46. [54]

    Im- agenet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Im- agenet: A large-scale hierarchical image database,” in 2009 IEEE conference on computer vision and pattern recognition. Ieee, 2009, pp. 248–255

    2009

  47. [55]

    Adam: A method for stochastic optimiza- tion,

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

    2014 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.