Pith. sign in

REVIEW 4 major objections 5 minor 30 references

Improving Equity in Health Modeling with GPT4-Turbo Generated Synthetic Data: A Comparative Study

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper finds that augmenting minority-group training data with GPT-4-Turbo–generated synthetic tabular records improves downstream model performance in 13 of 17 AUROC comparisons, but the benefit is inconsistent and group-specific…

desk verdict Honest, useful empirical comparison of LLM synthetic data augmentation for health equity, but the headline win rate rests on tiny test sets and missing confidence intervals. read the letter →

arxiv 2412.16335 v1 pith:2VBNAESY submitted 2024-12-20 cs.LG cs.CY

Daniel Smolyak , Arshana Welivita , Margrét V. Bjarnadóttir , Ritu Agarwal This is my paper · ORCID
classification cs.LGcs.CY
keywords largelanguagemodelssyntheticdatahealthequitytabularaugmentationunderrepresentedgroupsMIMIC-IVFraminghamHeartStudy
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 asks whether synthetic tabular data generated by a large language model can improve predictive performance for underrepresented racial groups in health datasets, where models typically do worse because fewer samples exist. It builds a pipeline that prompts GPT-4-Turbo with 20 real examples from a minority group and uses the returned synthetic records to augment the training set for a downstream logistic regression. Across two datasets (MIMIC-IV and Framingham) and three minority groups, the augmented models beat standard baselines in 13 of 17 AUROC comparisons, with gains up to 0.11 AUROC for Hispanic participants in Framingham. However, the improvement is inconsistent, and explicitly naming the group in the prompt adds little over a generic prompt. The authors conclude the approach is a complementary tool, not a guaranteed fix, for subgroup performance gaps.

What carries the argument

The central mechanism is a prompt-based synthetic data generation pipeline that follows the structure of Seedat et al. [26]: the model receives a role description, dataset context, 20 randomly selected real examples from the target minority group (resampled to include at least one positive outcome), and instructions to produce diverse new samples. A variant adds an explicit group label (e.g., 'generate data for 10 Asian patients') to test whether group specification matters. The generated records are concatenated with the real training data, and a logistic regression with a group indicator is trained and evaluated per group, compared against upweighting, separate models, and SMOTE as baselines.

What would settle it

Compute bootstrap confidence intervals for the AUROC differences between GPT-4-Turbo augmentation and the baseline for the Framingham minority groups; if most intervals include zero, the claim of superior augmentation is not supported. Alternatively, a replication using a larger held-out test set or an external cohort that finds no consistent improvement would falsify the central claim.

Watch

Extended reading notes

Core claim

The central claim is that LLM-generated synthetic data can serve as a useful augmentation tool for improving prediction performance on underrepresented demographic groups in tabular health data, but it is not uniformly superior. In a majority of experiments, models trained with GPT-4-Turbo–augmented data achieved higher AUROC and AUPRC than the baseline and than standard approaches such as upweighting, separate models, and SMOTE; the largest AUROC gain was 0.1121 (17%) for Hispanic participants in Framingham on the CHD outcome. Yet the paper finds no consistent benefit from making the prompt group-specific: a generic prompt that does not name the racial group often performs as well or better. The authors also report that GPT-4-Turbo's synthetic data captures some but not all of the real data's correlation structure, and that it tends to fall between minority and majority groups in racial-classifier probability space. They frame the method as 'another tool in the toolbox' for health-equity modeling rather than a standalone solution.

Load-bearing premise

The evaluation assumes that the held-out test sets for minority groups—especially the roughly 111 Asian participants in Framingham—are large enough for the reported AUROC differences to be reliable rather than dominated by sampling noise.

Editorial extensions

If this is right

  • Healthcare modelers can use out-of-the-box LLM generation as a low-effort augmentation option for small subgroups without fine-tuning or training a custom generator.
  • If the mixed results hold, practitioners should evaluate augmentation effects per dataset, group, and outcome rather than assuming universal benefit.
  • The small difference between generic and group-specific prompting suggests that prompt engineering for demographic labels may not be a productive lever for tabular health data.
  • The method can be combined with existing fairness interventions, such as upweighting or separate models, for cases where LLM augmentation alone underperforms.
  • The observed improvements, though often modest, can be clinically meaningful in low-prevalence settings where AUROC gains of 0.1 are substantial.

Reading between the lines

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

  • The negligible difference between generic and race-tailored prompts suggests that GPT-4-Turbo's pretraining already encodes demographic distributions in the feature space, making explicit labels redundant; a testable extension is to compare prompts that also provide group-specific medical relationships from external literature.
  • The tiny test sets for Framingham minority groups (e.g., 111 Asian participants) imply that reported AUROC differences of a few hundredths may fall within sampling noise; larger validation cohorts or bootstrap confidence intervals are needed before relying on specific effect sizes.
  • The method's success appears tied to how separable the minority group's feature-outcome distribution is from the majority's; extending the analysis to other protected attributes, such as age or socioeconomic status, could reveal where augmentation helps most.
  • The correlation between synthetic and real data being weaker for some pairwise relationships (e.g., between visit types in MIMIC-IV) hints that improving prompt context with explicit causal or clinical knowledge could make generation more faithful and, potentially, more beneficial.
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

4 major / 5 minor

Summary. The paper proposes using GPT4-Turbo (GPT4T) to generate synthetic tabular data for underrepresented racial/ethnic groups, with prompts that either do or do not name the target group, and evaluates whether augmenting the minority training data with these synthetic samples improves downstream logistic-regression AUROC/AUPRC for that group. Experiments are run on MIMIC-IV and Framingham data for Asian, Black, and Hispanic subgroups, with 25 repeated training samples per setting, and are compared against baseline, upweighting, separate models, and SMOTE. The authors report that GPT4T augmentation outperforms the standard baselines in 13 of 17 AUROC comparisons (9 of 17 AUPRC), that group-specific prompting gives little consistent benefit over a generic prompt, and that the method should be viewed as one tool among several for improving subgroup performance.

Significance. If the empirical claims were statistically supported, the paper would be a useful practical contribution: it evaluates an out-of-the-box LLM augmentation strategy on two widely used health datasets, compares against several standard baselines, reports exploratory quality checks, and provides detailed appendices with the prompt template, temperature sensitivity, and minority-size sensitivity analyses. The comparative study design is appropriate for the research question, and the cautious discussion is more measured than the abstract. However, the headline result that GPT4T augmentation is 'generally superior' rests on point estimates from very small held-out test sets without confidence intervals or significance tests, so the central claim is not yet established at the level of precision the paper claims.

major comments (4)
  1. [Section 2.5 and Table S1] The evaluation is underpowered for the Framingham minority groups, and the test-set construction is ambiguous. Table S1 reports only 111 Asian participants in Framingham; Section 2.3 removes 100 minority individuals for training and Section 2.2 supplies 20 minority examples to the prompt, while Section 2.5 says the hold-out set is all remaining samples after removing both the prompt examples and the training samples. Under a natural reading, the Asian test set is at most 11 patients, and if the prompt examples are disjoint from the training sample it could be even smaller. With an 11.7% CHD prevalence, this yields roughly one positive test case, so a single patient can move the AUROC by a large amount. Nevertheless, Table 2 reports four-decimal AUROC differences (e.g., 0.8139 vs 0.8130 for Asian hospitalization). Since the test set is fixed, averaging over 25 training samples reduces only sampling variation in the training data, not the test-set sampling variance. The paper itself excludes Asian CHF because of only 3 positive cases, but the same caution is not applied to Asian CHD/CVD, where the test sets are similarly tiny. Please report bootstrap confidence intervals or significance tests for every cell, or explicitly relabel the table as point estimates not suitable for ranking methods.
  2. [Section 3.2 and Table 2] The central '13 out of 17' superiority count is not robust because many wins are numerically tiny. Examples include MIMIC-IV Asian hospitalization (0.8139 vs 0.8130, a difference of 0.0009) and multiple generic-vs-tailored comparisons reported as 'no additional benefit' with differences below 0.01. Without confidence intervals or paired significance tests, these point estimates are indistinguishable from noise, and the win count could easily flip under resampling. The conclusion that GPT4T augmentation 'generally outperforms' baselines should be softened unless the authors can show that the relevant differences are statistically reliable, including the null result that group-specific prompting provides little benefit, which currently reads as accepting the null hypothesis without equivalence bounds.
  3. [Appendix D] The temperature hyperparameter is tuned on MIMIC-IV only, and the choice of 0.9 is then transferred to Framingham with the statement 'we expect similar results on the Framingham dataset.' This transfer is load-bearing because the main Table 2 includes all Framingham outcomes. The reported MIMIC-IV insensitivity (AUROC within 1% across temperatures) reduces the risk, but it does not establish that the same holds for Framingham, especially given different feature distributions and outcome prevalences. Please either report a Framingham temperature sensitivity analysis or state this transfer as an explicit limitation.
  4. [Section 2.4 and Table 2] The comparison across methods treats each of the 17 settings as an independent test, but the settings share data, preprocessing, and the LLM generation pipeline, so the 13/17 and 9/17 counts are not independent Bernoulli trials. A formal analysis accounting for clustering by dataset/outcome, or at least a per-setting significance test, is needed before drawing aggregate conclusions such as 'in the majority of experiments our method outperforms standard modeling baselines.'
minor comments (5)
  1. [Table 2] The abbreviation footnote says 'CHF: coronary heart failure,' but the standard term is 'congestive heart failure'; please correct this.
  2. [Throughout] The paper alternates between 'GPT4-Turbo' and 'GPT4T'; please define the abbreviation once at first use and use it consistently thereafter.
  3. [Section 2.5] Please clarify explicitly whether the 20 prompt examples in Section 2.2 are a subset of the 100 training patients or are drawn separately; the current description of the hold-out set does not make this clear.
  4. [Figure 4] The density plots are difficult to distinguish in grayscale; using different line styles or hatching would improve readability.
  5. [Appendix D] The temperature table header formats 'T emp' instead of 'Temperature'; please fix the typo.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's claims are empirical benchmark comparisons with disclosed tuning, not derivations that reduce to their inputs.

full rationale

This paper makes empirical claims about whether GPT-4Turbo-generated synthetic data improves subgroup AUROC/AUPRC in downstream logistic regression. The central comparison (Table 2) is an experimental evaluation against defined baselines; there is no equation-level derivation in which a predicted quantity is defined in terms of the outcome it is supposed to predict. The only element resembling tuning is the choice of temperature 0.9 in Appendix D, which was selected using downstream AUROC on MIMIC-IV. This is not circular because the appendix reports the full temperature comparison and states that AUROC scores at 0.9 are 'always within 1% of the scores for the other temperatures,' so the main results do not depend on a fitted parameter that was optimized to force a particular outcome. The paper also explicitly reports mixed results, including cases where baselines win, which is inconsistent with a claim forced by construction. Citations such as Seedat et al. for prompt structure and Movva et al. for a pre-processing pipeline are prior-work adaptations, not self-citations carrying the argument. The concern that Framingham Asian hold-out sets contain about 11 patients is a statistical power and uncertainty issue about small-sample AUROC estimates, not a circularity issue. No load-bearing step reduces to its own input by definition or by self-citation, so the appropriate circularity score is 0.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The paper's empirical claim rests on a small set of design choices (temperature, prompt-example count, minority sample size) and domain assumptions about the validity of LLM-generated tabular data and of AUROC as an equity measure. None of these are provided with independent evidence; the temperature choice in particular is fitted on the same evaluation task.

free parameters (3)
  • LLM sampling temperature = 0.9
    Temperature for GPT4-Turbo chosen in Appendix D to optimize downstream AUROC on MIMIC-IV; 0.9 selected as compromise between realism and diversity.
  • Number of prompt examples = 20
    Fixed in Section 2.2 with resampling to ensure at least one positive outcome; not varied in main experiments.
  • Minority group training sample size = 100
    Chosen in Section 2.3; sensitivity analysis (Appendix E) varies 50/100/200 on MIMIC-IV only.
assumptions (5)
  • domain assumption GPT4-Turbo can generate tabular samples that are valid additions to real training data for a logistic regression model.
    Core premise of the augmentation pipeline; invoked throughout Sections 2.2 and 2.4.
  • domain assumption AUROC/AUPRC measured on held-out real data is an adequate proxy for health-equity-relevant model performance.
    Section 2.5; the paper equates improved minority AUROC with improved equity in the title and abstract.
  • domain assumption Coarse racial/ethnic categories (Asian, Black, Hispanic, White) are meaningful stable group definitions.
    Section 2.1; all analysis conditions on these labels.
  • ad hoc to paper Temperature tuned on MIMIC-IV transfers to Framingham without re-tuning.
    Appendix D: 'we expect similar results on the Framingham dataset'.
  • domain assumption Logistic regression with a binary group indicator is an appropriate base model for subgroup comparisons.
    Section 2.4; the model is not varied, yet the conclusions are about model performance generally.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Improving Equity in Health Modeling with GPT4-Turbo Generated Synthetic Data: A Comparative Study." pith.science (2026). https://pith.science/paper/2VBNAESY

@misc{pith2026241216335,
  author       = {Pith},
  title        = {Pith review of: Improving Equity in Health Modeling with GPT4-Turbo Generated Synthetic Data: A Comparative Study},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2VBNAESY}},
  note         = {Machine review of arXiv:2412.16335}
}
read the original abstract

Objective. Demographic groups are often represented at different rates in medical datasets. These differences can create bias in machine learning algorithms, with higher levels of performance for better-represented groups. One promising solution to this problem is to generate synthetic data to mitigate potential adverse effects of non-representative data sets. Methods. We build on recent advances in LLM-based synthetic data generation to create a pipeline where the synthetic data is generated separately for each demographic group. We conduct our study using MIMIC-IV and Framingham "Offspring and OMNI-1 Cohorts" datasets. We prompt GPT4-Turbo to create group-specific data, providing training examples and the dataset context. An exploratory analysis is conducted to ascertain the quality of the generated data. We then evaluate the utility of the synthetic data for augmentation of a training dataset in a downstream machine learning task, focusing specifically on model performance metrics across groups. Results. The performance of GPT4-Turbo augmentation is generally superior but not always. In the majority of experiments our method outperforms standard modeling baselines, however, prompting GPT-4-Turbo to produce data specific to a group provides little to no additional benefit over a prompt that does not specify the group. Conclusion. We developed a method for using LLMs out-of-the-box to synthesize group-specific data to address imbalances in demographic representation in medical datasets. As another "tool in the toolbox", this method can improve model fairness and thus health equity. More research is needed to understand the conditions under which LLM generated synthetic data is useful for non-representative medical data sets.

Figures

Figures reproduced from arXiv: 2412.16335 by the authors.

Figure 1
Figure 1. Pipeline overview: first the data from the designated majority group and minor [PITH_FULL_IMAGE:figures/full_fig_p010_1.png] view at source ↗
Figure 2
Figure 2. (left) Kernel density estimation for the joint distribution of diastolic and sys [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. (left) MIMIC-IV: Distribution of the distances between each GPT4T generated [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Density of predicted probabilities for belonging to the minority racial group for [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 28 canonical work pages

  1. [1]

    Mitigating racial biases for machine learning based skin cancer detection

    Julian Abhari and Ashwin Ashok. Mitigating racial biases for machine learning based skin cancer detection. In Proceedings of the Twenty- fourth International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing , pages 556–561, 2023

    work page 2023

  2. [2]

    Predicting with proxies: Transfer learning in high di- mension

    Hamsa Bastani. Predicting with proxies: Transfer learning in high di- mension. Management Science, 67(5):2964–2984, 2021

    work page 2021

  3. [3]

    Comparison of machine learning algorithms for clinical event prediction (risk of coronary heart disease)

    Juan-Jose Beunza, Enrique Puertas, Ester Garc ´ ıa-Ovejero, Gema Vil- lalba, Emilia Condes, Gergana Koleva, Cristian Hurtado, and Manuel F Landecho. Comparison of machine learning algorithms for clinical event prediction (risk of coronary heart disease). Journal of biomedical infor- matics, 97:103257, 2019

    work page 2019

  4. [4]

    Language Models are Realistic Tabular Data Gen- erators

    Vadim Borisov, Kathrin Sessler, Tobias Leemann, Martin Pawelczyk, and Gjergji Kasneci. Language Models are Realistic Tabular Data Gen- erators. September 2022

    work page 2022

  5. [5]

    Smote: synthetic minority over-sampling technique

    Nitesh V Chawla, Kevin W Bowyer, Lawrence O Hall, and W Philip Kegelmeyer. Smote: synthetic minority over-sampling technique. Jour- nal of artificial intelligence research , 16:321–357, 2002. 17

    work page 2002

  6. [6]

    Why is my classi- fier discriminatory? Advances in neural information processing systems, 31, 2018

    Irene Chen, Fredrik D Johansson, and David Sontag. Why is my classi- fier discriminatory? Advances in neural information processing systems, 31, 2018

    work page 2018

  7. [7]

    Ethical machine learning in healthcare

    Irene Y Chen, Emma Pierson, Sherri Rose, Shalmali Joshi, Kadija Fer- ryman, and Marzyeh Ghassemi. Ethical machine learning in healthcare. Annual review of biomedical data science , 4:123–144, 2021

    work page 2021

  8. [8]

    Benchmarking large language models in retrieval-augmented generation

    Jiawei Chen, Hongyu Lin, Xianpei Han, and Le Sun. Benchmarking large language models in retrieval-augmented generation. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 38, pages 17754–17762, 2024

    work page 2024

Show all 30 references
  1. [9]

    A review of medical image data aug- mentation techniques for deep learning applications

    Phillip Chlap, Hang Min, Nym Vandenberg, Jason Dowling, Lois Hol- loway, and Annette Haworth. A review of medical image data aug- mentation techniques for deep learning applications. Journal of Medical Imaging and Radiation Oncology , 65(5):545–563, 2021

    2021

  2. [10]

    Scaling instruction-finetuned language models

    Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Yunxuan Li, Xuezhi Wang, Mostafa Dehghani, Sid- dhartha Brahma, et al. Scaling instruction-finetuned language models. Journal of Machine Learning Research , 25(70):1–53, 2024

    2024

  3. [11]

    PreciseDebias: An Automatic Prompt Engineering Approach for Generative AI to Mitigate Image Demographic Biases

    Colton Clemmer, Junhua Ding, and Yunhe Feng. PreciseDebias: An Automatic Prompt Engineering Approach for Generative AI to Mitigate Image Demographic Biases. In 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) , pages 8581–8590, Waikoloa, HI, USA, Janua...

    2024

  4. [12]

    Symptoms across the continuum of acute coronary syndromes: dif- ferences between women and men

    Holli A DeVon, Catherine J Ryan, Amy L Ochs, and Moshe Shapiro. Symptoms across the continuum of acute coronary syndromes: dif- ferences between women and men. American journal of critical care , 17(1):14–24, 2008

    2008

  5. [13]

    The joint lasso: high-dimensional regression for group structured data

    Frank Dondelinger, Sach Mukherjee, and Alzheimer’s Disease Neu- roimaging Initiative. The joint lasso: high-dimensional regression for group structured data. Biostatistics, 21(2):219–235, 2020

    2020

  6. [14]

    Race/ethnic differences in the associations of the framingham risk factors with carotid imt and cardiovascular events

    Crystel M Gijsberts, Karlijn A Groenewegen, Imo E Hoefer, Mari- nus JC Eijkemans, Folkert W Asselbergs, Todd J Anderson, Annie R 18 Britton, Jacqueline M Dekker, Gunnar Engstr¨ om, Greg W Evans, et al. Race/ethnic differences in the associations of the framingham risk factors ...

    2015

  7. [15]

    Evaluation and Mitigation of Racial Bias in Clinical Machine Learning Models: Scoping Review

    Jonathan Huang, Galal Galal, Mozziyar Etemadi, and Mahesh Vaidyanathan. Evaluation and Mitigation of Racial Bias in Clinical Machine Learning Models: Scoping Review. JMIR Medical Informatics, 10(5):e36388, May 2022

    2022

  8. [16]

    Mimic-iv

    Alistair Johnson, Lucas Bulgarelli, Tom Pollard, Steven Horng, Leo An- thony Celi, and Roger Mark. Mimic-iv. PhysioNet. Available online at: https://physionet. org/content/mimiciv/1.0/(accessed August 23, 2021), pages 49–55, 2020

    2021

  9. [17]

    Assessing socioeconomic bias in machine learning algorithms in health care: a case study of the HOUSES index

    Young J Juhn, Euijung Ryu, Chung-Il Wi, Katherine S King, Momin Malik, Santiago Romero-Brufau, Chunhua Weng, Sunghwan Sohn, Richard R Sharp, and John D Halamka. Assessing socioeconomic bias in machine learning algorithms in health care: a case study of the HOUSES index. Journa...

    2022

  10. [18]

    Group-wise prompting for synthetic tabular data generation using large language models

    Jinhee Kim, Taesung Kim, and Jaegul Choo. Group-wise prompting for synthetic tabular data generation using large language models. arXiv preprint arXiv:2404.12404, 2024

    2024 arXiv

  11. [19]

    Diaz, Sai Batchu, and Brandon Lucke- Wold

    Giona Kleinberg, Michael J. Diaz, Sai Batchu, and Brandon Lucke- Wold. Racial underrepresentation in dermatological datasets leads to biased machine learning models and inequitable healthcare. Journal of biomed research, 3(1):42, 2022

    2022

  12. [20]

    Generative models improve fair- ness of medical classifiers under distribution shifts

    Ira Ktena, Olivia Wiles, Isabela Albuquerque, Sylvestre-Alvise Rebuffi, Ryutaro Tanno, Abhijit Guha Roy, Shekoofeh Azizi, Danielle Belgrave, Pushmeet Kohli, Taylan Cemgil, et al. Generative models improve fair- ness of medical classifiers under distribution shifts. Nature Medi...

    2024

  13. [21]

    A survey on bias and fairness in machine learning

    Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. A survey on bias and fairness in machine learning. ACM computing surveys (CSUR) , 54(6):1–35, 2021. 19

    2021

  14. [22]

    Coarse race data conceals dispar- ities in clinical risk score performance, August 2023

    Rajiv Movva, Divya Shanmugam, Kaihua Hou, Priya Pathak, John Gut- tag, Nikhil Garg, and Emma Pierson. Coarse race data conceals dispar- ities in clinical risk score performance, August 2023

    2023

  15. [23]

    Synthetic data generation: State of the art in health care domain

    Hajra Murtaza, Musharif Ahmed, Naurin Farooq Khan, Ghulam Mur- taza, Saad Zafar, and Ambreen Bano. Synthetic data generation: State of the art in health care domain. Computer Science Review, 48:100546, May 2023

    2023

  16. [24]

    Enrollment of racial minorities in clinical trials: old problem assumes new urgency in the age of immunotherapy

    Bassel Nazha, Manoj Mishra, Rebecca Pentz, and Taofeek K Owonikoko. Enrollment of racial minorities in clinical trials: old problem assumes new urgency in the age of immunotherapy. American Society of Clinical Oncology Educational Book, 39:3–10, 2019

    2019

  17. [25]

    Yoonyoung Park, Jianying Hu, Moninder Singh, Issa Sylla, Irene Dankwa-Mullan, Eileen Koski, and Amar K. Das. Comparison of Meth- ods to Reduce Bias From Clinical Prediction Models of Postpartum Depression. JAMA Network Open , 4(4):e213909, April 2021

    2021

  18. [26]

    Curated LLM: Synergy of LLMs and Data Curation for tabular augmentation in ultra low-data regimes, February 2024

    Nabeel Seedat, Nicolas Huynh, Boris van Breugel, and Mihaela van der Schaar. Curated LLM: Synergy of LLMs and Data Curation for tabular augmentation in ultra low-data regimes, February 2024

    2024

  19. [27]

    Towards expert-level medical question answering with large lan- guage models

    Karan Singhal, Tao Tu, Juraj Gottweis, Rory Sayres, Ellery Wulczyn, Le Hou, Kevin Clark, Stephen Pfohl, Heather Cole-Lewis, Darlene Neal, et al. Towards expert-level medical question answering with large lan- guage models. arXiv preprint arXiv:2305.09617 , 2023

    2023 arXiv

  20. [28]

    Cohort profile: The framingham heart study (fhs): overview of milestones in cardiovascular epidemiology

    Connie W Tsao and Ramachandran S Vasan. Cohort profile: The framingham heart study (fhs): overview of milestones in cardiovascular epidemiology. International journal of epidemiology , 44(6):1800–1813, 2015

    2015

  21. [29]

    Fairness without harm: De- coupled classifiers with preference guarantees

    Berk Ustun, Yang Liu, and David Parkes. Fairness without harm: De- coupled classifiers with preference guarantees. In International Confer- ence on Machine Learning , pages 6373–6382. PMLR, 2019

    2019

  22. [30]

    hospital admissions and readmission

    Travis Zack, Eric Lehman, Mirac Suzgun, Jorge A. Rodriguez, Leo An- thony Celi, Judy Gichoya, Dan Jurafsky, Peter Szolovits, David W. Bates, Raja-Elie E. Abdulnour, Atul J. Butte, and Emily Alsentzer. 20 Assessing the potential of GPT-4 to perpetuate racial and gender biases i...

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.