pith. sign in

arxiv: 2605.29744 · v1 · pith:FB2SCDBEnew · submitted 2026-05-28 · 💻 cs.AI · cs.CL· cs.LG· cs.MA

Why Specialist Models Still Matter: A Heterogeneous Multi-Agent Paradigm for Medical Artificial Intelligence

Yanan Wang , Shuaicong Hu , Jian Liu , Guohui Zhou , Aiguo Wang , Cuiwei Yang This is my paper

Pith reviewed 2026-06-29 07:41 UTC · model grok-4.3

classification 💻 cs.AI cs.CLcs.LGcs.MA
keywords medical AImulti-agent systemslarge language modelsspecialist modelsclinical decision makingheterogeneous collaboration
0
0 comments X

The pith

Specialist models remain essential because their collaboration with generalist LLMs outperforms either type alone on clinical tasks.

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

The paper claims that generalist large language models will not make domain-specific medical specialist models obsolete. Instead, the future of medical AI depends on orchestrating collaboration among generalist LLMs, specialist models, and clinicians through a multi-agent system. This matters because it preserves the precision of specialists in handling specific data modalities while leveraging the broad reasoning of general models. The proposed framework introduces mechanisms for fusing conflicting evidence and deciding when to involve human clinicians. Experiments on three real clinical decision tasks support that the combined approach delivers better results than isolated use of either model type.

Core claim

The central claim is that a heterogeneous multi-agent framework called HetMedAgent, which coordinates generalist LLMs with domain-specific specialist models and clinicians through conflict-aware evidence fusion, uncertainty-triggered intervention, and adaptive calibration, produces significantly higher performance on three real-world clinical decision-making tasks than using generalist LLMs or specialist models in isolation.

What carries the argument

HetMedAgent, a heterogeneous medical multi-agent framework that enables conflict-aware evidence fusion, uncertainty-based clinician intervention triggering, and adaptive threshold calibration.

If this is right

  • Specialist models hold irreplaceable value for modality-specific medical analysis.
  • Medical AI development should move from monolithic foundation models toward multi-agent collaboration.
  • The approach balances broad reasoning from generalist models with the precision of specialists.
  • Clinician intervention can be triggered selectively based on model uncertainty.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Medical AI systems may benefit from explicit interfaces that let different model types exchange evidence rather than competing as replacements.
  • The same collaboration pattern could be tested in other high-stakes domains where general and specialized tools coexist.
  • Future work could examine whether the performance edge persists when specialist models are updated independently of the generalist LLM.

Load-bearing premise

The three chosen real-world clinical tasks and the selected specialist models are representative enough for performance gains to be credited to the collaboration rather than to task-specific details or implementation choices.

What would settle it

Running the same framework on a fourth independent clinical task where the combined system shows no advantage over the stronger of the two model types used alone.

Figures

Figures reproduced from arXiv: 2605.29744 by Aiguo Wang, Cuiwei Yang, Guohui Zhou, Jian Liu, Shuaicong Hu, Yanan Wang.

Figure 1
Figure 1. Figure 1: (a) Previous: The generalist medical LLM. Lacking domain specificity and human oversight, single LLMs are prone to hallucinations and unsafe decisions on multimodal data. (b) Ours: Overview of the HetMedAgent framework. Orchestrates a Specialist Agent Group, a Generalist LLM Agent, and a Clinician Agent, enabling conflict-aware evidence fusion and uncertainty-driven oversight with adaptive calibration. act… view at source ↗
Figure 2
Figure 2. Figure 2: Detailed architecture of the HetMedAgent system. (Left) Specialist models convert multimodal data into standardized text. (Top-Right) Orchestration and reasoning workflow. (Bottom-Right) Shared memory module. (Thirunavukarasu et al., 2023; Lee et al., 2023). Medi￾cal LLMs like Med-PaLM (Singhal et al., 2023) improve through fine-tuning on medical data (Saab et al., 2024; Yang et al., 2022) but face high tr… view at source ↗
Figure 3
Figure 3. Figure 3: HetMedAgent’s complete decision-making workflow with inputs and outputs. tency rather than lexical overlap. Weighted Evidence Fusion. The system receives outputs {F w 1 , F w 2 , . . . , F w k } from all specialist agents and com￾putes integration weights to resolve conflicts. We assemble the weighted evidence into a single reasoning input via a deterministic evidence assembly protocol: Inputreason = Conte… view at source ↗
Figure 5
Figure 5. Figure 5: (Left) Uncertainty decomposition analysis (fixed thresh￾old θP = 0.5). (Right) Comparing F1 scores between au￾tonomous decisions and cases requiring clinician intervention across clinical decision-making tasks (fixed threshold θP = 0.5; Mann–Whitney U test, p < 0.001). θP = 0.5 Adaptive θP 0 25 50 75 100 125 Intervention count 114 97 -17 cases Risk stratification Etiology Severity 1.4 1.5 1.6 1.7 1.8 AIR 1… view at source ↗
Figure 4
Figure 4. Figure 4: Performance comparison between weighted and direct evidence fusion strategies across clinical decision-making tasks. Clinical Decision-Making Tasks: evaluating admission risk stratification, etiology prediction, and severity assess￾ment using AUROC and F1-Score. Both Transformer￾based specialist models output diagnostic text with confi￾dence scores. In our implementation, we use GPT-4o as the generalist LL… view at source ↗
Figure 7
Figure 7. Figure 7: Case study of HetMedAgent’s autonomous decision-making from multimodal inputs to final clinical decisions. HetMedAgent achieves average improvements of 6.6% in AUROC and 7.9% in F1 score over the best single-model baseline (Meditron), and 4.3% in AUROC and 5.7% in F1 score over the best multi-agent system (MedAgents). Our Transformer-based specialist models (Aw ECHO and Aw ECG) substantially outperform tra… view at source ↗
Figure 8
Figure 8. Figure 8: Structured prompt ψtask used by the orchestrator agent. 21 [PITH_FULL_IMAGE:figures/full_fig_p021_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Structured prompt ψreason used by the reasoning agent. The prompt instructs the LLM to reason with clinical knowledge and medical guidelines, and generate preliminary decisions with explicit reasoning chains. 22 [PITH_FULL_IMAGE:figures/full_fig_p022_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Controlled extreme examples for cross-specialist conflict detection. Case (Top): semantically equivalent findings expressed with different terminology. Case (Bottom): findings with partial lexical overlap but contradictory semantics. emb(·): semantic embed￾ding via PubMedBERT bi-encoder. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Demographic distributions of the multimodal test set. (Left) Gender distribution (unique patients). (Right) Age distribution by gender (all cases) [PITH_FULL_IMAGE:figures/full_fig_p024_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Modal ablation experiments for HetMedAgent (w/o Clinician) showing task-level F1 and AUC scores across different modality configurations. ECHO: echocardiography only; ECG: electrocardiogram only; ECHO+ECG: multimodal fusion. 24 [PITH_FULL_IMAGE:figures/full_fig_p024_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Complete routing trace for the representative autonomous case study. The visualization expands [PITH_FULL_IMAGE:figures/full_fig_p025_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Subgroup analysis of HetMedAgent (w/o Clinician) on the cardiovascular clinical test set, stratified by sex (Male/Female) and age group (<65, 65–74, 75–84, ≥85). ∗ p < 0.05; ∗∗ p < 0.01; ∗ ∗ ∗ p < 0.001; ns: not significant (Fisher’s exact test) [PITH_FULL_IMAGE:figures/full_fig_p026_14.png] view at source ↗
read the original abstract

The impressive performance of generalist large language models (LLMs) such as GPT and Claude in healthcare raises a critical question: will domain-specific medical specialist models become obsolete? We argue that the future of medical artificial intelligence (AI) lies not in building monolithic medical foundation models, nor in replacing human expertise, but in orchestrating collaboration among generalist LLMs, domain-specific specialist models, and clinicians. We propose HetMedAgent, a heterogeneous medical multi-agent framework that enables conflict-aware evidence fusion, uncertainty-based clinician intervention triggering, and adaptive threshold calibration. Experiments on three real-world clinical decision-making tasks demonstrate that the synergy between generalist LLMs and domain-specific specialist models significantly outperforms using either type of model alone, validating the irreplaceable value of specialist models in modality-specific analysis. HetMedAgent represents a shift from building medical LLMs or foundation models to multi-agent collaboration, achieving a balance between general reasoning capabilities and domain-specific precision.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper argues that specialist models remain essential in medical AI and proposes HetMedAgent, a heterogeneous multi-agent framework that orchestrates generalist LLMs, domain-specific specialist models, and clinicians via conflict-aware evidence fusion, uncertainty-based intervention triggering, and adaptive threshold calibration. It claims that experiments on three real-world clinical decision-making tasks demonstrate that this synergy significantly outperforms using either model type alone, validating the irreplaceable role of specialists in modality-specific analysis.

Significance. If the reported performance gains are robustly demonstrated with proper controls, the work could support a shift from monolithic medical foundation models toward multi-agent collaboration paradigms that combine general reasoning with domain precision. The proposed mechanisms (conflict-aware fusion and uncertainty triggering) provide a concrete integration strategy worth further exploration in clinical settings.

major comments (1)
  1. [Abstract] Abstract: The central claim that 'the synergy ... significantly outperforms using either type of model alone' on three tasks supplies no task definitions, specialist model choices, baselines (generalist-only, specialist-only, or simple ensembles), metrics, statistical tests, or ablation controls. This absence makes it impossible to attribute gains to the proposed conflict-aware fusion or uncertainty triggering rather than task artifacts or implementation details, rendering the validation of specialist models' value untestable.
minor comments (1)
  1. [Abstract] The abstract introduces the term 'HetMedAgent' and its components without a high-level architectural diagram or pseudocode that would clarify the interaction flow among agents.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for highlighting the need for greater specificity in the abstract. We address this point below and will revise accordingly.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that 'the synergy ... significantly outperforms using either type of model alone' on three tasks supplies no task definitions, specialist model choices, baselines (generalist-only, specialist-only, or simple ensembles), metrics, statistical tests, or ablation controls. This absence makes it impossible to attribute gains to the proposed conflict-aware fusion or uncertainty triggering rather than task artifacts or implementation details, rendering the validation of specialist models' value untestable.

    Authors: We agree the abstract is high-level and omits these specifics due to length constraints. The full manuscript defines the three clinical tasks, specifies the specialist models used for each modality, details the baselines (generalist-only, specialist-only, and simple ensembles), reports metrics with statistical tests, and includes ablations on the fusion and triggering components. We will revise the abstract to include concise references to tasks, key metrics, and controls to make the claim more testable from the abstract alone. revision: yes

Circularity Check

0 steps flagged

No circularity in derivation chain

full rationale

The paper proposes HetMedAgent as a multi-agent framework and supports its value via experimental results on three clinical tasks. No equations, fitted parameters, self-referential derivations, or load-bearing self-citations appear in the provided text. The central claim rests on empirical outperformance rather than any mathematical reduction to inputs by construction, making the derivation self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

Review performed on abstract only; ledger entries are therefore limited to claims visible in the abstract text.

free parameters (1)
  • adaptive threshold calibration parameters
    Mentioned as part of the framework but no values or fitting procedure given in abstract.
axioms (1)
  • domain assumption Specialist models supply modality-specific analysis that generalist LLMs cannot replicate at equivalent precision.
    This premise underpins the entire argument that specialist models remain irreplaceable.
invented entities (1)
  • HetMedAgent no independent evidence
    purpose: Heterogeneous multi-agent orchestration layer with conflict-aware fusion and uncertainty triggering.
    Newly named system introduced in the abstract; no independent evidence supplied.

pith-pipeline@v0.9.1-grok · 5714 in / 1300 out tokens · 27192 ms · 2026-06-29T07:41:53.020692+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

51 extracted references · 12 canonical work pages · 9 internal anchors

  1. [1]

    GPT-4 Technical Report

    Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F. L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  2. [2]

    The claude 3 model family: Opus, sonnet, haiku

    Anthropic AI . The claude 3 model family: Opus, sonnet, haiku. Claude-3 Model Card, 1 0 (1): 0 4, 2024

    2024

  3. [3]

    MEDITRON-70B: Scaling Medical Pretraining for Large Language Models

    Chen, Z., Cano, A. H., Romanou, A., Bonnet, A., Matoba, K., Salvi, F., Pagliardini, M., Fan, S., K \"o pf, A., Mohtashami, A., et al. Meditron-70b: Scaling medical pretraining for large language models. arXiv preprint arXiv:2311.16079, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  4. [4]

    Cohen, I. G. and Mello, M. M. Hipaa and protecting health information in the 21st century. JAMA, 320 0 (3): 0 231--232, 2018

    2018

  5. [5]

    S., and Jurdak, R

    Dorri, A., Kanhere, S. S., and Jurdak, R. Multi-agent systems: A survey. IEEE Access, 6: 0 28573--28593, 2018

    2018

  6. [6]

    A., Ko, J., Swetter, S

    Esteva, A., Kuprel, B., Novoa, R. A., Ko, J., Swetter, S. M., Blau, H. M., and Thrun, S. Dermatologist-level classification of skin cancer with deep neural networks. Nature, 542 0 (7639): 0 115--118, 2017

    2017

  7. [7]

    Ethical and legal challenges of artificial intelligence-driven healthcare

    Gerke, S., Minssen, T., and Cohen, G. Ethical and legal challenges of artificial intelligence-driven healthcare. In Artificial intelligence in healthcare, pp.\ 295--336. Elsevier, 2020

    2020

  8. [8]

    Ghassemi, M., Oakden-Rayner, L., and Beam, A. L. The false hope of current approaches to explainable artificial intelligence in health care. The Lancet Digital Health, 3 0 (11): 0 e745--e750, 2021

    2021

  9. [9]

    Domain-specific language model pretraining for biomedical natural language processing

    Gu, Y., Tinn, R., Cheng, H., Lucas, M., Usuyama, N., Liu, X., Naumann, T., Gao, J., and Poon, H. Domain-specific language model pretraining for biomedical natural language processing. ACM Transactions on Computing for Healthcare (HEALTH), 3 0 (1): 0 1--23, 2021

    2021

  10. [10]

    C., Wu, D., Narayanaswamy, A., Venugopalan, S., Widner, K., Madams, T., Cuadros, J., et al

    Gulshan, V., Peng, L., Coram, M., Stumpe, M. C., Wu, D., Narayanaswamy, A., Venugopalan, S., Widner, K., Madams, T., Cuadros, J., et al. Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA, 316 0 (22): 0 2402--2410, 2016

    2016

  11. [11]

    DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

    Guo, D., Yang, D., Zhang, H., Song, J., Zhang, R., Xu, R., Zhu, Q., Ma, S., Wang, P., Bi, X., et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  12. [12]

    Y., Rajpurkar, P., Haghpanahi, M., Tison, G

    Hannun, A. Y., Rajpurkar, P., Haghpanahi, M., Tison, G. H., Bourn, C., Turakhia, M. P., and Ng, A. Y. Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network. Nature Medicine, 25 0 (1): 0 65--69, 2019

    2019

  13. [13]

    Multimodal integration in health care: development with applications in disease management

    Hao, Y., Cheng, C., Li, J., Li, H., Di, X., Zeng, X., Jin, S., Han, X., Liu, C., Wang, Q., et al. Multimodal integration in health care: development with applications in disease management. Journal of medical Internet research, 27: 0 e76557, 2025

    2025

  14. [14]

    Deep residual learning for image recognition

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

    2016

  15. [15]

    Metagpt: Meta programming for a multi-agent collaborative framework

    Hong, S., Zhuge, M., Chen, J., Zheng, X., Cheng, Y., Wang, J., Zhang, C., Yau, S., Lin, Z., et al. Metagpt: Meta programming for a multi-agent collaborative framework. In International Conference on Learning Representations, volume 2024, pp.\ 23247--23275, 2024

    2024

  16. [16]

    Transparent artificial intelligence-enabled interpretable and interactive sleep apnea assessment across flexible monitoring scenarios

    Hu, S., Liu, J., Wang, Y., Fu, C., Zhu, J., Yu, H., and Yang, C. Transparent artificial intelligence-enabled interpretable and interactive sleep apnea assessment across flexible monitoring scenarios. Nature Communications, 16 0 (1): 0 7548, 2025

    2025

  17. [17]

    H., and Quinn, G

    Inkpen, K., Chappidi, S., Mallari, K., Nushi, B., Ramesh, D., Michelucci, P., Mandava, V., Vep r ek, L. H., and Quinn, G. Advancing human-ai complementarity: The impact of user expertise and algorithmic tuning on joint decision making. ACM Transactions on Computer-Human Interaction, 30 0 (5): 0 1--29, 2023

    2023

  18. [18]

    S., Kazerooni, E

    Jabbour, S., Fouhey, D., Shepard, S., Valley, T. S., Kazerooni, E. A., Banovic, N., Wiens, J., and Sjoding, M. W. Measuring the impact of ai in the diagnosis of hospitalized patients: a randomized clinical vignette survey study. JAMA, 330 0 (23): 0 2275--2284, 2023

    2023

  19. [19]

    F., McCoy Jr, T

    Jacobs, M., Pradier, M. F., McCoy Jr, T. H., Perlis, R. H., Doshi-Velez, F., and Gajos, K. Z. How machine-learning recommendations influence clinician treatment selections: the example of antidepressant selection. Translational Psychiatry, 11 0 (1): 0 108, 2021

    2021

  20. [20]

    Mistral 7B

    Jiang, A. Q., Sablayrolles, A., Mensch, A., et al. Mistral 7b. arXiv preprint arXiv:2310.06825, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  21. [21]

    Biomistral: A collection of open-source pretrained large language models for medical domains

    Labrak, Y., Bazoge, A., Morin, E., Gourraud, P.-A., Rouvier, M., and Dufour, R. Biomistral: A collection of open-source pretrained large language models for medical domains. arXiv preprint arXiv:2402.10373, 2024

    work page arXiv 2024

  22. [22]

    W., Brown, K

    Lamb, B. W., Brown, K. F., Nagpal, K., Vincent, C., Green, J. S., and Sevdalis, N. Quality of care management decisions by multidisciplinary cancer teams: a systematic review. Annals of surgical oncology, 18 0 (8): 0 2116--2125, 2011

    2011

  23. [23]

    H., and Kang, J

    Lee, J., Yoon, W., Kim, S., Kim, D., Kim, S., So, C. H., and Kang, J. Biobert: a pre-trained biomedical language representation model for biomedical text mining. Bioinformatics, 36 0 (4): 0 1234--1240, 2020

    2020

  24. [24]

    Benefits, limits, and risks of gpt-4 as an ai chatbot for medicine

    Lee, P., Bubeck, S., and Petro, J. Benefits, limits, and risks of gpt-4 as an ai chatbot for medicine. New England Journal of Medicine, 388 0 (13): 0 1233--1239, 2023

    2023

  25. [25]

    E., Motzfeldt, A

    Li \'e vin, V., Hother, C. E., Motzfeldt, A. G., and Winther, O. Can large language models reason about medical questions? Patterns, 5 0 (3), 2024

    2024

  26. [26]

    Decision making strategies and team efficacy in human-ai teams

    Munyaka, I., Ashktorab, Z., Dugan, C., Johnson, J., and Pan, Q. Decision making strategies and team efficacy in human-ai teams. Proceedings of the ACM on Human-Computer Interaction, 7 0 (CSCW1): 0 1--24, 2023

    2023

  27. [27]

    Capabilities of GPT-4 on Medical Challenge Problems

    Nori, H., King, N., McKinney, S. M., Carignan, D., and Horvitz, E. Capabilities of gpt-4 on medical challenge problems. arXiv preprint arXiv:2303.13375, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  28. [28]

    Rajpurkar, P., Chen, E., Banerjee, O., and Topol, E. J. Ai in health and medicine. Nature Medicine, 28 0 (1): 0 31--38, 2022

    2022

  29. [29]

    Evaluation framework to guide implementation of ai systems into healthcare settings

    Reddy, S., Rogers, W., Makinen, V.-P., Coiera, E., Brown, P., Wenzel, M., Weicken, E., Ansari, S., Mathur, P., Casey, A., et al. Evaluation framework to guide implementation of ai systems into healthcare settings. BMJ Health & Care Informatics, 28 0 (1): 0 e100444, 2021

    2021

  30. [30]

    Capabilities of Gemini Models in Medicine

    Saab, K., Tu, T., Weng, W.-H., Tanno, R., Stutz, D., Wulczyn, E., Zhang, F., Strother, T., Park, C., Vedadi, E., et al. Capabilities of gemini models in medicine. arXiv preprint arXiv:2404.18416, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  31. [31]

    Llama-3-meditron: An open-weight suite of medical llms based on llama-3.1

    Sallinen, A., Solergibert, A.-J., Zhang, M., Boy \'e , G., Dupont-Roc, M., Theimer-Lienhard, X., Boisson, E., Bernath, B., Hadhri, H., Tran, A., et al. Llama-3-meditron: An open-weight suite of medical llms based on llama-3.1. In Workshop on Large Language Models and Generative AI for Health at AAAI 2025, 2025

    2025

  32. [32]

    AgentClinic: a multimodal agent benchmark to evaluate AI in simulated clinical environments

    Schmidgall, S., Ziaei, R., Harris, C., Reis, E., Jopling, J., and Moor, M. Agentclinic: a multimodal agent benchmark to evaluate ai in simulated clinical environments. arXiv preprint arXiv:2405.07960, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  33. [33]

    R., Vassef, S., Goyal, A., Kumar, N., and Saha, K

    Shimgekar, S. R., Vassef, S., Goyal, A., Kumar, N., and Saha, K. Agentic ai framework for end-to-end medical data inference. arXiv preprint arXiv:2507.18115, 2025

    work page arXiv 2025

  34. [34]

    Large language models encode clinical knowledge

    Singhal, K., Azizi, S., Tu, T., et al. Large language models encode clinical knowledge. Nature, 620 0 (7972): 0 172--180, 2023

    2023

  35. [35]

    Quo vadis, ai-empowered doctor? JMIR medical education, 11 0 (1): 0 e70079, 2025

    Takahashi, G., von Liechti, L., and Tarshizi, E. Quo vadis, ai-empowered doctor? JMIR medical education, 11 0 (1): 0 e70079, 2025

    2025

  36. [36]

    L., Tatekawa, H., Saito, K., Tsujimoto, Y., Miki, Y., and Ueda, D

    Takita, H., Kabata, D., Walston, S. L., Tatekawa, H., Saito, K., Tsujimoto, Y., Miki, Y., and Ueda, D. A systematic review and meta-analysis of diagnostic performance comparison between generative ai and physicians. npj Digital Medicine, 8 0 (1): 0 175, 2025

    2025

  37. [37]

    and Le, Q

    Tan, M. and Le, Q. Efficientnet: Rethinking model scaling for convolutional neural networks. In International conference on machine learning, pp.\ 6105--6114. PMLR, 2019

    2019

  38. [38]

    Medagents: Large language models as collaborators for zero-shot medical reasoning

    Tang, X., Zou, A., Zhang, Z., Li, Z., Zhao, Y., Zhang, X., Cohan, A., and Gerstein, M. Medagents: Large language models as collaborators for zero-shot medical reasoning. In Findings of the Association for Computational Linguistics: ACL 2024, pp.\ 599--621, 2024

    2024

  39. [39]

    J., Ting, D

    Thirunavukarasu, A. J., Ting, D. S. J., Elangovan, K., Gutierrez, L., Tan, T. F., and Ting, D. S. W. Large language models in medicine. Nature Medicine, 29 0 (8): 0 1930--1940, 2023

    1930

  40. [40]

    D., and Goldenberg, A

    Tonekaboni, S., Joshi, S., McCradden, M. D., and Goldenberg, A. What clinicians want: contextualizing explainable machine learning for clinical end use. In Machine Learning for Healthcare Conference, pp.\ 359--380. PMLR, 2019

    2019

  41. [41]

    Deep medicine: how artificial intelligence can make healthcare human again

    Topol, E. Deep medicine: how artificial intelligence can make healthcare human again. Hachette UK, 2019

    2019

  42. [42]

    G., Schalekamp, S., Rutten, M

    van Leeuwen, K. G., Schalekamp, S., Rutten, M. J., van Ginneken, B., and de Rooij, M. Artificial intelligence in radiology: 100 commercially available products and their scientific evidence. European Radiology, 31 0 (6): 0 3797--3804, 2021

    2021

  43. [43]

    and Von dem Bussche, A

    Voigt, P. and Von dem Bussche, A. The eu general data protection regulation (gdpr). A practical guide, 1st ed., Cham: Springer International Publishing, 10 0 (3152676): 0 10--5555, 2017

    2017

  44. [44]

    V., Zhou, D., et al

    Wei, J., Wang, X., Schuurmans, D., Bosma, M., Xia, F., Chi, E., Le, Q. V., Zhou, D., et al. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35: 0 24824--24837, 2022

    2022

  45. [45]

    Pmc-llama: toward building open-source language models for medicine

    Wu, C., Lin, W., Zhang, X., Zhang, Y., Xie, W., and Wang, Y. Pmc-llama: toward building open-source language models for medicine. Journal of the American Medical Informatics Association, 31 0 (9): 0 1833--1843, 2024

    2024

  46. [46]

    AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation

    Wu, Q., Bansal, G., Zhang, J., Wu, Y., Zhang, S., Zhu, E., Li, B., Jiang, L., Zhang, X., and Wang, C. Autogen: Enabling next-gen llm applications via multi-agent conversation framework. arXiv preprint arXiv:2308.08155, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  47. [47]

    Doctorglm: Fine-tuning your chinese doctor is not a herculean task

    Xiong, H., Wang, S., Zhu, Y., Zhao, Z., Liu, Y., Huang, L., Wang, Q., and Shen, D. Doctorglm: Fine-tuning your chinese doctor is not a herculean task. arXiv preprint arXiv:2304.01097, 2023

    work page arXiv 2023

  48. [48]

    C., Smith, K

    Yang, X., Chen, A., PourNejatian, N., Shin, H. C., Smith, K. E., Parisien, C., Compas, C., Martin, C., Costa, A. B., Flores, M. G., et al. A large language model for electronic health records. NPJ Digital Medicine, 5 0 (1): 0 194, 2022

    2022

  49. [49]

    Tree of thoughts: Deliberate problem solving with large language models

    Yao, S., Yu, D., Zhao, J., Shafran, I., Griffiths, T., Cao, Y., and Narasimhan, K. Tree of thoughts: Deliberate problem solving with large language models. Advances in neural information processing systems, 36: 0 11809--11822, 2023

    2023

  50. [50]

    BiomedCLIP: a multimodal biomedical foundation model pretrained from fifteen million scientific image-text pairs

    Zhang, S., Xu, Y., Usuyama, N., Bagga, J., Tinn, R., Preston, S., Rao, R., Wei, M., Valluri, N., Wong, C., et al. Large-scale domain-specific pretraining for biomedical vision-language processing. arXiv preprint arXiv:2303.00915, 2 0 (3): 0 6, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  51. [51]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...