Pith. sign in

REVIEW 4 major objections 5 minor 44 references

This paper claims to be the first to study physician intent trajectories in doctor-patient dialogues, releasing a 5,292-turn SOAP-labeled corpus and showing that current models classify single intents well but predict next intents poorly.

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

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

A new SOAP-based physician intent taxonomy and labeled dialogue dataset, with benchmarks showing models classify intents accurately but fail to predict SOAP-category transitions.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A genuinely useful SOAP-aligned intent dataset with a solid taxonomy, but the trajectory benchmark numbers are suspect because the train/test split is done at the turn level, so same-dialogue context almost certainly leaks. the 4 major comments →

arxiv 2508.19077 v1 pith:QZQZN2G5 submitted 2025-08-26 cs.CL

"Where does it hurt?" -- Dataset and Study on Physician Intent Trajectories in Doctor Patient Dialogues

Tom R\"ohr , Soumyadeep Roy , Fares Al Mohamad , Jens-Michalis Papaioannou , Wolfgang Nejdl , Felix Gers , Alexander L\"oser This is my paper
classification cs.CL
keywords physician intent trajectoriesSOAP taxonomydoctor-patient dialoguemedical intent classificationnext intent predictiondialogue summarizationclinical NLP dataset
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

This paper tries to establish that doctor-patient dialogues have a measurable structure -- the sequence of physician intents -- and that this structure can be captured with a taxonomy built on the SOAP note framework. The authors annotate more than 5,000 turns from 207 role-played clinical conversations with 20 fine-grained physician intents, verify the labels with medical experts, and release the corpus with guidelines. Benchmarking shows that a fine-tuned clinical encoder classifies a single doctor utterance's intent well (0.69 macro average precision) but predicts the next intent far worse (0.37), indicating that current models understand the content of medical dialogue better than its trajectory. The paper also reports common trajectory patterns and shows that filtering dialogues by predicted intent before summarization improves generated notes for most sections. If correct, this provides a reusable resource and a specific diagnosis of where medical dialogue systems need improvement.

Core claim

The central discovery is that a clinical conversation can be read as a trajectory through SOAP phases, and that current models learn the phases but not the transitions. The paper builds a 20-intent taxonomy with clinicians, annotates 5,292 doctor-patient turns from the Aci-bench role-played dialogues, and verifies labels with crowd-sourced medical experts (81.13% agreement). Fine-tuned GatorTronS reaches macro-AP 0.69 on intent classification but only 0.37 on next-intent prediction; reconstruction experiments show it misses SOAP phase boundaries, predicting transitions late or early and defaulting to linear paths on non-linear dialogues. The paper concludes models understand dialogue structu

What carries the argument

The machinery is the SOAP intent trajectory: each doctor utterance is labeled with one or more of 20 intents nested under the clinical note categories Subjective, Objective, Assessment, and Plan. This taxonomy converts unstructured dialogue into a discrete state sequence, making trajectory analysis and prediction possible. The next-intent task feeds up to five preceding doctor-patient turns into the same fine-tuned encoders, and a hierarchical fine-tuning variant first predicts the SOAP category and then masks intents outside that category; comparing the two tasks is what exposes the transition bottleneck.

Load-bearing premise

The load-bearing premise is that the role-played Aci-bench dialogues stand in for real doctor-patient consultations, a limitation the paper states in its final section; the next-intent numbers additionally assume that the stratified train/test splits do not carry turns from the same dialogue into both sides.

What would settle it

Re-split the 5,292 turns by dialogue so all turns of one Aci-bench conversation stay in the same fold, fine-tune GatorTronS on next-intent prediction, and compare macro-AP with the reported 0.37. A large drop would show the transition numbers were inflated by within-dialogue leakage; a small drop would confirm the transition-failure result.

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

If this is right

  • A public 5,292-turn, 20-intent corpus with annotation guidelines gives clinical NLP a shared test bed, so future physician-intent models can be compared on the same trajectories.
  • Because intent classification (0.69 macro-AP) clearly outperforms next-intent prediction (0.37 macro-AP), practical decision support should use intent classifiers for structuring and coding conversations, not for anticipating the doctor's next move.
  • Common trajectory patterns -- Subjective-dominant openings, brief Assessment phases, repeated Plan negotiation loops, chitchat scattered throughout -- give differential-diagnosis systems a prior to plan against.
  • Intent filtering before summarization improves average decoder-only summarization by 5.39 percent, with the largest gain in objective-exam sections, so intent-aware preprocessing is a cheap, effective step for clinical note generation.
  • The model's linear-default behavior on non-linear dialogues means evaluations of medical dialogue systems should include follow-up and annual-exam conversations, not only first-visit complaints.

Where Pith is reading between the lines

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

  • The corpus is naturally suited to a phase-boundary detection task: predict the turn where the doctor switches SOAP categories. The paper's error analysis suggests this is where models fail, so a dedicated transition model could improve downstream trajectory reconstruction.
  • Because Lab Examination is often confused with Radiology and Physical Examination, adding a test-type entity-recognition auxiliary loss would likely sharpen both intent classification and the summarization filter.
  • A testable extension: rank consultations by the model's transition uncertainty and filter only the uncertain ones before summarization; the paper's verbosity results predict that non-linear and chitchat-heavy dialogues would gain most.
  • Applying the same annotation protocol to non-role-played or de-identified real clinical dialogues would test how much the trajectory patterns and filtering gains transfer beyond synthetic consultations.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces a physician-intent annotation dataset built on the Aci-bench doctor-patient dialogues. With input from medical professionals, the authors define a 20-class taxonomy organized under SOAP categories, annotate or verify more than 5,000 turns via Prolific, and release the data, guidelines, and code. They benchmark encoder and decoder-only models on two tasks: classifying the intent of a single physician utterance, and predicting the intent of the next physician turn given up to five preceding turns. They also analyze common SOAP trajectories, evaluate how well a fine-tuned model reconstructs dialogue sequences, and test whether intent-based filtering improves medical-dialogue summarization. The main reported findings are that fine-tuned clinical encoders (especially GatorTronS) perform well on intent classification but poorly on next-intent prediction, that models fail to detect SOAP phase transitions, and that filtering improves summarization for several settings.

Significance. If the claims hold, the main contribution is a reusable, SOAP-aligned physician-intent corpus with annotation guidelines, which would be of clear value to clinical dialogue research. The public release of the dataset is a concrete strength, as is the effort to involve medical professionals in taxonomy design and verification. The trajectory analysis is also potentially useful for designing diagnostic dialogue systems. However, the evaluation claims rest on experiment-design choices that are currently not fully controlled; in particular, the lack of dialogue-level separation between training and test splits and the overlap between the summarization filter's training data and the summarized dialogues could materially change the reported numbers. The central dataset contribution is not invalidated, but the benchmarking and downstream-filtering conclusions require re-validation with cleaner protocols.

major comments (4)
  1. [Section 4.1, Tables 2-3] The train/validation/test splits are described only as 'stratified sampling' over the 5,292 labeled turns, with no statement that all turns from one Aci-bench dialogue are kept in the same split. Aci-bench contains only 207 dialogues with roughly 27 turns each, so turn-level random splitting almost certainly places turns from the same dialogue in both training and test. For next-intent prediction, whose input is up to five preceding doctor-patient turns, the model can then memorize a specific dialogue's trajectory from nearby training turns, inflating the reported GatorTronS 0.85 AUROC / 0.37 AP and undermining the Section 5.3 transition-error analysis. Please re-run the benchmarks using dialogue-level grouped splits and report the number of dialogues in each split, or explicitly justify why turn-level splits are valid.
  2. [Section 6] The intent filter is the GatorTronS model fine-tuned on the 5,292 labeled turns from the same Aci-bench dialogues that are later filtered and summarized. This creates a circular evaluation: the filter has already seen these dialogues during training, so the reported 'significant boost' from filtering may be optimistic and may not transfer to unseen dialogues. The summarization experiments should be repeated with a filter trained on a dialogue-disjoint set, or at minimum the paper should report results for filtering with a model trained only on held-out dialogues. This is essential to support the abstract's claim that intent filtering improves summarization.
  3. [Section 3.1] The data verification is summarized by a single accuracy figure of 81.13%, with the remaining 19.87% reviewed and 'unresolved disagreements' removed. It is not stated how many annotators labeled each turn, what inter-annotator agreement was, how accuracy was computed (before or after removing unresolved disagreements), or how many samples were removed and according to what criteria. Because the dataset is the paper's central contribution, this level of detail is insufficient. Please report full annotation statistics, including per-class precision/recall against the in-house labels, Cohen's kappa or similar, and the exact exclusion rule.
  4. [Section 5.3] The conclusion that 'the model does not learn phase transitions' is supported only by qualitative inspection of four reconstructed sequences in Figure 5. The procedure for generating the 'Model Output' sequences from the next-intent-prediction model is not fully specified, and no automatic metric quantifies transition errors over the full test set. Since this analysis is a central piece of the paper's main claim that models fail to identify transitions between SOAP categories, please define the reconstruction algorithm and report quantitative transition-error statistics (e.g., precision/recall for SOAP-category changes, timing errors) on the dialogue-level test split.
minor comments (5)
  1. [Table 1] Typo: '1,77' should be '1.77'.
  2. [Figure 5] The token-level sequence plots are nearly unreadable at printed size; consider enlarging fonts, using banded SOAP category colors, or separating into panels.
  3. [Section 6] The summarization setup does not state whether the Aci-bench train/test split from Yim et al. is used; please clarify the split and how the filter's training dialogues relate to the summarization test dialogues.
  4. [Section 6] The phrase 'significant boost' is not backed by statistical significance testing. Report confidence intervals or paired tests for the filter comparisons, especially where gains are small (e.g., GPT-4o assessment-and-plan: 0.50 to 0.51).
  5. [Section 5.1] The percentage differences, e.g., 'AP drops by 114.89%' for hierarchical fine-tuning, are confusing because the drop is from 0.37 to 0.10, which is a percentage-point reduction of 0.27; please report percentage-point differences or absolute changes.

Circularity Check

0 steps flagged

No significant circularity: the dataset, taxonomy, benchmarks, and summarization experiment are self-contained rather than derived from their own outputs.

full rationale

The paper's central contributions are (1) a SOAP-based intent taxonomy and a newly annotated corpus, (2) benchmark results for intent classification and next-intent prediction, and (3) a downstream experiment using intent filtering for summarization. None of these reduces to its inputs by construction. The taxonomy is derived from the external SOAP standard and human annotation, not from the model outputs; the classification and next-intent benchmarks are standard supervised evaluations on an external dataset (Aci-bench); and the summarization experiment compares filtered vs. unfiltered inputs using the same Aci-bench reference summaries. The closest concern is that the summarization filter is trained on the same Aci-bench dialogues that are later filtered, which could overstate the benefit. However, this is a data-contamination/optimism issue, not a definitional circularity: the filter is trained on intent labels, not on summarization targets, and the paper actually reports mixed effects (e.g., filtering hurts BART on full-note and subjective summarization). Similarly, the turn-level stratified split without dialogue grouping could inflate next-intent prediction through same-dialogue leakage, but that is an experimental-validity threat, not a circular derivation. The paper's only self-citations are to its own supplementary material and a prior model reference; neither is load-bearing. Therefore no circular step is established under the required standard.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 0 invented entities

The paper's claims rest on three domain assumptions: the representativeness of role-played Aci-bench dialogues, the reliability of crowd-sourced expert labels, and the suitability of the SOAP schema for dialogue intents. It also implicitly assumes turn-level splitting prevents dialogue-level leakage. No new physical or mathematical entities are introduced.

axioms (4)
  • domain assumption Aci-bench role-played dialogues are representative of real doctor-patient interactions.
    Section 3.1 selects Aci-bench for annotation; the Limitation section states the dialogues are role-played and may not reflect real-world scenarios, yet the trajectory claims in Sections 3.2 and 5.3 generalize from them.
  • domain assumption Prolific-recruited medical experts produce reliable intent labels.
    The entire dataset and downstream benchmarks rest on the 81.13% verification accuracy; the qualification and screening of the roughly 90 Prolific workers is deferred to the supplementary material.
  • domain assumption SOAP is an appropriate schema for capturing physician intents in dialogue.
    Section 3.1 adopts SOAP because of clinical adoption, but the mapping of turns to 20 intents is a design choice that determines all trajectory statistics and transition analyses.
  • ad hoc to paper Turn-level random splits do not leak information from the same dialogue across train and test.
    Section 4.1 describes stratified sampling without specifying a dialogue-level split; next-intent memory of a specific dialogue could inflate results.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of "Where does it hurt?" -- Dataset and Study on Physician Intent Trajectories in Doctor Patient Dialogues." pith.science (2026). https://pith.science/paper/QZQZN2G5

@misc{pith2026250819077,
  author       = {Pith},
  title        = {Pith review of: "Where does it hurt?" -- Dataset and Study on Physician Intent Trajectories in Doctor Patient Dialogues},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QZQZN2G5}},
  note         = {Machine review of arXiv:2508.19077}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

In a doctor-patient dialogue, the primary objective of physicians is to diagnose patients and propose a treatment plan. Medical doctors guide these conversations through targeted questioning to efficiently gather the information required to provide the best possible outcomes for patients. To the best of our knowledge, this is the first work that studies physician intent trajectories in doctor-patient dialogues. We use the `Ambient Clinical Intelligence Benchmark' (Aci-bench) dataset for our study. We collaborate with medical professionals to develop a fine-grained taxonomy of physician intents based on the SOAP framework (Subjective, Objective, Assessment, and Plan). We then conduct a large-scale annotation effort to label over 5000 doctor-patient turns with the help of a large number of medical experts recruited using Prolific, a popular crowd-sourcing platform. This large labeled dataset is an important resource contribution that we use for benchmarking the state-of-the-art generative and encoder models for medical intent classification tasks. Our findings show that our models understand the general structure of medical dialogues with high accuracy, but often fail to identify transitions between SOAP categories. We also report for the first time common trajectories in medical dialogue structures that provide valuable insights for designing `differential diagnosis' systems. Finally, we extensively study the impact of intent filtering for medical dialogue summarization and observe a significant boost in performance. We make the codes and data, including annotation guidelines, publicly available at https://github.com/DATEXIS/medical-intent-classification.

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

44 extracted references · 29 canonical work pages · 1 internal anchor

  1. [1]

    Abdin, J

    M. Abdin, J. Aneja, H. Behl, et al. Phi-4 technical report, 2024. URL https://arxiv.org/abs/2412.08905

    Pith/arXiv arXiv 2024

  2. [2]

    Ben Abacha, W.-w

    A. Ben Abacha, W.-w. Yim, Y . Fan, et al. An empirical study of clinical note generation from doctor-patient encounters. In Proceed- ings of the 17th Conference of the European Chapter of the Asso- ciation for Computational Linguistics , pages 2291–2302, Dubrovnik, Croatia, May 2023. Association for Computational Linguistics. doi: 10.18653/v1/2023.eacl-main.168

    doi:10.18653/v1/2023.eacl-main.168 2023

  3. [4]

    S. Budd, T. Day, J. Simpson, et al. Can non-specialists pro- vide high quality gold standard labels in challenging modalities? In Domain Adaptation and Representation Transfer, and Afford- able Healthcare and AI for Resource Diverse Global Health , pages 251–262, Cham, 2021. Springer International Publishing. ISBN 978-3-030-87722-4. URL https://link.sprin...

    work page 2021

  4. [5]

    A. Chen, Z. Yu, X. Yang, et al. Contextualized medication informa- tion extraction using transformer-based deep learning architectures. J. Biomed. Inform., 142(104370):104370, June 2023. URL https://www. sciencedirect.com/science/article/pii/S1532046423000916

    work page 2023

  5. [6]

    W. Chen, Z. Li, H. Fang, et al. A benchmark for automatic medical consultation system: frameworks, tasks and datasets. Bioinformatics, 39, 2022. URL https://api.semanticscholar.org/CorpusID:248239674

    work page 2022

  6. [7]

    Cocos, T

    A. Cocos, T. Qian, C. Callison-Burch, et al. Crowd control: Effec- tively utilizing unscreened crowd workers for biomedical data annota- tion. Journal of Biomedical Informatics, 69:86–92, 2017. ISSN 1532-

    work page 2017

  7. [8]

    Dowlagar and R

    S. Dowlagar and R. Mamidi. A code-mixed task-oriented dialog dataset for medical domain. Computer Speech & Language, 78:101449, 2023. ISSN 0885-2308. doi: https://doi.org/10.1016/j.csl.2022.101449

    work page arXiv 2023

  8. [9]

    Enarvi, M

    S. Enarvi, M. Amoia, M. Del-Agua Teba, et al. Generating medical reports from patient-doctor conversations using sequence-to-sequence models. In Proceedings of the First Workshop on Natural Language Processing for Medical Conversations, pages 22–30, Online, July 2020. Association for Computational Linguistics. doi: 10.18653/v1/2020. nlpmc-1.4

    doi:10.18653/v1/2020 2020

  9. [10]

    D. Fast, L. C. Adams, F. Busch, et al. Autonomous medical evalu- ation for guideline adherence of large language models. npj Digital Medicine, 7(1):358, Dec. 2024. URL https://www.nature.com/articles/ s41746-024-01356-6

    work page 2024

  10. [11]

    Figueroa, J

    A. Figueroa, J. Papaioannou, C. Fallon, et al. Boosting long- tail data classification with sparse prototypical networks. In Ma- chine Learning and Knowledge Discovery in Databases. Research Track - European Conference, ECML PKDD 2024, Vilnius, Lithua- nia, September 9-13, 2024, Proceedings, Part VII , volume 14947 of Lecture Notes in Computer Science , p...

    work page doi:10.1007/978-3-031-70368-3 2024

  11. [12]

    Finley, W

    G. Finley, W. Salloum, N. Sadoughi, et al. From dictations to clinical reports using machine translation. In Proceedings of the 2018 Confer- ence of the North American Chapter of the Association for Computa- tional Linguistics: Human Language Technologies, Volume 3 (Industry Papers), pages 121–128, New Orleans - Louisiana, June 2018. Associ- ation for Com...

    work page doi:10.18653/v1/n18-3015 2018

  12. [13]

    Grattafiori, A

    A. Grattafiori, A. Dubey, A. Jauhri, et al. The llama 3 herd of models,

    work page

  13. [14]

    Y . Gu, R. Tinn, H. Cheng, et al. Domain-specific language model pre- training for biomedical natural language processing.ACM Transactions on Computing for Healthcare, 3(1):1–23, Oct. 2021. ISSN 2637-8051. doi: 10.1145/3458754

    doi:10.1145/3458754 2021

  14. [15]

    Y . Kim, C. Park, H. Jeong, et al. Mdagents: An adap- tive collaboration of llms for medical decision-making. In Advances in Neural Information Processing Systems , vol- ume 37, pages 79410–79452. Curran Associates, Inc., 2024. URL https://proceedings.neurips.cc/paper_files/paper/2024/file/ 90d1fc07f46e31387978b88e7e057a31-Paper-Conference.pdf

    work page 2024

  15. [16]

    Krishna, S

    K. Krishna, S. Khosla, J. Bigham, et al. Generating SOAP notes from doctor-patient conversations using modular summarization techniques. In Proceedings of the 59th Annual Meeting of the Association for Com- putational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) , pages 4958– 4972, Online, ...

    doi:10.18653/v1/2021.acl-long.384 2021

  16. [17]

    Lewis, Y

    M. Lewis, Y . Liu, N. Goyal, et al. BART: Denoising sequence-to- sequence pre-training for natural language generation, translation, and comprehension. In Proceedings of the 58th Annual Meeting of the Asso- ciation for Computational Linguistics , pages 7871–7880, Online, July

    work page

  17. [18]

    C.-Y . Lin. ROUGE: A package for automatic evaluation of summaries. In Text Summarization Branches Out, pages 74–81, Barcelona, Spain, July 2004. Association for Computational Linguistics. URL https:// aclanthology.org/W04-1013/

    work page 2004

  18. [19]

    Loshchilov and F

    I. Loshchilov and F. Hutter. Decoupled weight decay regularization. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019 . OpenReview.net, 2019. URL https://openreview.net/forum?id=Bkg6RiCqY7

    2019

  19. [20]

    Michalopoulos, K

    G. Michalopoulos, K. Williams, G. Singh, et al. MedicalSum: A guided clinical abstractive summarization model for generating medical re- ports from patient-doctor conversations. In Findings of the Associa- tion for Computational Linguistics: EMNLP 2022 , pages 4741–4749, Abu Dhabi, United Arab Emirates, Dec. 2022. Association for Compu- tational Linguisti...

    work page doi:10.18653/v1/2022.findings-emnlp.349 2022

  20. [21]

    Papadopoulos Korfiatis, F

    A. Papadopoulos Korfiatis, F. Moramarco, R. Sarac, et al. PriMock57: A dataset of primary care mock consultations. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguis- tics (Volume 2: Short Papers) , pages 588–598, Dublin, Ireland, May

    work page

  21. [22]

    V . Parsons. Stratified Sampling. 02 2017. ISBN 9781118445112. doi: 10.1002/9781118445112.stat05999.pub2

    work page doi:10.1002/9781118445112.stat05999.pub2 2017

  22. [23]

    work of reference unique in medical literature

    J. Pearn. Herbert french (1875-1951) and his differential diagno- sis a “work of reference unique in medical literature”. J. Med. Bi- ogr., 30(2):131–135, May 2022. URL https://pubmed.ncbi.nlm.nih.gov/ 32954933/

    work page 1951

  23. [24]

    Rajchl, L

    M. Rajchl, L. M. Koch, C. Ledig, et al. Employing weak annotations for medical image analysis problems. CoRR, abs/1708.06297, 2017

    Pith/arXiv arXiv 2017

  24. [25]

    S. E. Robertson, S. Walker, S. Jones, et al. Okapi at TREC-3. In Pro- ceedings of The Third Text REtrieval Conference, TREC 1994, Gaithers- burg, Maryland, USA, November 2-4, 1994 , volume 500-225 of NIST Special Publication , pages 109–126. National Institute of Standards and Technology (NIST), 1994. URL https://dblp.org/rec/conf/trec/ RobertsonWJHG94

    work page 1994

  25. [26]

    T. Röhr, A. Figueroa, J.-M. Papaioannou, et al. Revisiting clinical out- come prediction for MIMIC-IV. In Proceedings of the 6th Clinical Natural Language Processing Workshop, pages 208–217, Mexico City, Mexico, June 2024. Association for Computational Linguistics. doi: 10.18653/v1/2024.clinicalnlp-1.18

    work page doi:10.18653/v1/2024.clinicalnlp-1.18 2024

  26. [27]

    Rose, C.-C

    D. Rose, C.-C. Hung, M. Lepri, et al. Meddxagent: A unified modular agent framework for explainable automatic differential diagnosis, 2025. URL https://arxiv.org/abs/2502.19175

    Pith/arXiv arXiv 2025

  27. [28]

    where does it hurt?

    T. Röhr. (supplementary material) "where does it hurt?" - dataset and study on physician intent trajectories in doctor patient dialogues, 2025. URL https://doi.org/10.5281/zenodo.16941593

    work page doi:10.5281/zenodo.16941593 2025

  28. [29]

    V . V . Saley, G. Saha, R. J. Das, et al. MediTOD: An English dialogue dataset for medical history taking with comprehensive annotations. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 16843–16877, Miami, Florida, USA, Nov

    work page 2024

  29. [30]

    Soldaini and N

    L. Soldaini and N. Goharian. Quickumls: a fast, unsupervised approach for medical concept extraction. Special Interest Group on Information Retrieval, MedIR Workshop, 2016. URL https://ir.cs.georgetown.edu/ downloads/quickumls.pdf

    work page 2016

  30. [31]

    T. Tu, M. Schaekermann, A. Palepu, et al. Towards conversational di- agnostic artificial intelligence. Nature, 642(8067):442–450, June 2025. URL https://www.nature.com/articles/s41586-025-08866-7

    work page 2025

  31. [32]

    L. L. Weed. The problem oriented record as a basic tool in medical ed- ucation, patient care and clinical research. Annals of clinical research, 3(3):131–134, 1971. URL https://pubmed.ncbi.nlm.nih.gov/4934176/

    work page arXiv 1971

  32. [33]

    B. T. Willard and R. Louf. Efficient guided generation for llms. arXiv preprint arXiv:2307.09702 , 2023. URL https://arxiv.org/abs/ 2307.09702

    Pith/arXiv arXiv 2023

  33. [34]

    doi: 10.18653/v1/ 2024.emnlp-main.936

    Association for Computational Linguistics. doi: 10.18653/v1/ 2024.emnlp-main.936

    doi:10.18653/v1/ 2024

  34. [35]

    G. Yan, J. Pei, P. Ren, et al. Mˆ2-meddialog: A dataset and bench- marks for multi-domain multi-service medical dialogues. CoRR, abs/2109.00430, 2021. URL https://arxiv.org/abs/2109.00430

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  35. [36]

    A. Yang, B. Yang, B. Hui, et al. Qwen2 technical report, 2024. URL https://arxiv.org/abs/2407.10671

    Pith/arXiv arXiv 2024

  36. [37]

    X. Yang, N. PourNejatian, H. C. Shin, et al. Gatortron: A large clin- ical language model to unlock patient information from unstructured electronic health records. medRxiv, 2022. doi: 10.1101/2022.02.27. 22271257

    work page doi:10.1101/2022.02.27 2022

  37. [38]

    W.-w. Yim, Y . Fu, A. Ben Abacha, et al. Aci-bench: a novel ambient clinical intelligence dataset for benchmarking automatic visit note gen- eration. Scientific Data, 10(1):586, Sep 2023. ISSN 2052-4463. doi: 10.1038/s41597-023-02487-3

    doi:10.1038/s41597-023-02487-3 2023

  38. [39]

    Y . Wu, M. Jiang, J. Xu, et al. Clinical named entity recognition us- ing deep learning models. AMIA. In Annual Symposium proceedings. AMIA Symposium, pages 1812–1819. 2017. URL https://pubmed.ncbi. nlm.nih.gov/29854252/

    work page arXiv 2017

  39. [40]

    Zhang, R

    L. Zhang, R. Negrinho, A. Ghosh, et al. Leveraging pretrained mod- els for automatic summarization of doctor-patient conversations. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 3693–3712, Punta Cana, Dominican Republic, Nov. 2021. Association for Computational Linguistics. doi: 10.18653/v1/2021. findings-emnlp.313

    doi:10.18653/v1/2021 2021

  40. [41]

    Zhang*, V

    T. Zhang*, V . Kishore*, F. Wu*, K. Q. Weinberger, and Y . Artzi. Bertscore: Evaluating text generation with bert. In International Con- ference on Learning Representations , 2020. URL https://openreview. net/forum?id=SkeHuCVFDr

    work page 2020

  41. [42]

    Zhang, Z

    Y . Zhang, Z. Jiang, T. Zhang, et al. Mie: A medical information extrac- tor towards medical dialogues. InAnnual Meeting of the Association for Computational Linguistics, 2020. URL https://aclanthology.org/2020. acl-main.576/

    work page 2020

  42. [44]

    G. Zeng, W. Yang, Z. Ju, et al. MedDialog: Large-scale medical dia- logue datasets. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 9241–9250, Online, Nov. 2020. Association for Computational Linguistics. doi: 10.18653/v1/2020.emnlp-main.743

    doi:10.18653/v1/2020.emnlp-main.743 2020

  43. [464]

    doi: https://doi.org/10.1016/j.jbi.2017.04.003

    work page doi:10.1016/j.jbi.2017.04.003 2017

  44. [2024]

    URL https://arxiv.org/abs/2407.21783

    Pith/arXiv arXiv

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.