Comparing Two Model Designs for Clinical Note Generation; Is an LLM a Useful Evaluator of Consistency?
Open accessAbstract
Following an interaction with a patient, physicians are responsible for the submission of clinical documentation, often organized as a SOAP note.A clinical note is not simply a summary of the conversation but requires the use of appropriate medical terminology.The relevant information can then be extracted and organized according to the structure of the SOAP note.In this paper we analyze two different approaches to generate the different sections of a SOAP note based on the audio recording of the conversation, and specifically examine them in terms of note consistency.The first approach generates the sections independently, while the second method generates them all together.In this work we make use of PEGASUS-X Transformer models and observe that both methods lead to similar ROUGE values (less than 1% difference) and have no difference in terms of the Factuality metric.We perform a human evaluation to measure aspects of consistency and demonstrate that LLMs like Llama2 can be used to perform the same tasks with roughly the same agreement as the human annotators.Between the Llama2 analysis and the human reviewers we observe a Cohen Kappa inter-rater reliability of 0.79, 1.00, and 0.32 for consistency of age, gender, and body part injury, respectively.With this we demonstrate the usefulness of leveraging an LLM to measure quality indicators that can be identified by humans but are not currently captured by automatic metrics.This allows scaling evaluation to larger data sets, and we find that clinical note consistency improves by generating each new section conditioned on the output of all previously generated sections.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
A LLaMA-2 evaluator attains human-level agreement (Cohen's κ = 0.79) when checking the internal consistency of clinical notes.
“Brake and Schaaf [12] showed that a LLaMA-2 evaluator attains human-level agreement (Cohen's 𝜅 = 0 . 79) when checking the internal consistency of clinical notes”
From A Survey on LLM-as-a-Judge (2024) · cites this paper as Brake and Schaaf [12]
Table 7 reports Cohen’s κ = 0.79 between the 2-shot Llama2 evaluator and the consensus human choice for age consistency, matching the score of a single medical expert. The claim presents 0.79 as agreement for clinical-note consistency generally, but the reported agreement varies by criterion: κ = 1.00 for gender, 0.32 for body part, and 0.00 for coherence in that configuration. (LLM, conf=0.91)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
No references could be resolved yet.
Research works printed but not linked: 24
- Asma Ben Abacha, Wen-wai Yim, Griffin Adams, Neal Snider, and Meliha Yetisgen. 2023. Overview of the MEDIQA-chat 2023 shared tasks on the sum- marization & generation of doctor-patient conver- sations. In Proceedings of the 5th Clinical Natu- ral Language Processing Workshop, pages 503-513, Toronto, Canada. Association for Computational Lin- guistics. · DOI · Paper page
- Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Ma- teusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei. 2020. Language models are few-shot learners. In Ad- vances in Neural Information Processing Systems, volume 33, pages 1877-1901.
- Cheng-Han Chiang and Hung-yi Lee. 2023. Can large language models be an alternative to human evalua- tions? In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Vol- ume 1: Long Papers), pages 15607-15631, Toronto, Canada. Association for Computational Linguistics. · DOI · Paper page
- Fabrizio Gilardi, Meysam Alizadeh, and Maël Kubli. 2023. Chatgpt outperforms crowd workers for text-annotation tasks. Proceedings of the National Academy of Sciences, 120(30). · DOI · Paper page
- John Glover, Federico Fancellu, Vasudevan Jagan- nathan, Matthew R. Gormley, and Thomas Schaaf. 2022. Revisiting text decomposition methods for NLI-based factuality scoring of summaries. In Pro- ceedings of the 2nd Workshop on Natural Language Generation, Evaluation, and Metrics (GEM), pages 97-105, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computational Linguistics. · DOI
- Seungone Kim, Jamin Shin, Yejin Cho, Joel Jang, Shayne Longpre, Hwaran Lee, Sangdoo Yun, Seongjin Shin, Sungdong Kim, James Thorne, and Minjoon Seo. 2024. Prometheus: Inducing evalua- tion capability in language models. In The Twelfth International Conference on Learning Representa- tions.
- Kundan Krishna, Sopan Khosla, Jeffrey Bigham, and Zachary C. Lipton. 2021. Generating SOAP notes from doctor-patient conversations using modular summarization techniques. In Proceedings of the 59th Annual Meeting of the Association for Compu- tational Linguistics and the 11th International Joint Conference on Natural Language Processing (Vol- ume 1: Long Papers), pages 4958-4972, Online. As- sociation for Computational Linguistics. · DOI
- Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Veselin Stoyanov, and Luke Zettlemoyer. 2020. BART: Denoising sequence-to-sequence pre-training for natural language generation, translation, and com- prehension. In Proceedings of the 58th Annual Meet- ing of the Association for Computational Linguistics, pages 7871-7880, Online. Association for Computa- tional Linguistics. · DOI · Paper page
- Chin-Yew Lin. 2004. ROUGE: A package for auto- matic evaluation of summaries. In Text Summariza- tion Branches Out, pages 74-81, Barcelona, Spain. Association for Computational Linguistics.
- Yang Liu, Dan Iter, Yichong Xu, Shuohang Wang, Ruochen Xu, and Chenguang Zhu. 2023. G-eval: NLG evaluation using gpt-4 with better human align- ment. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 2511-2522, Singapore. Association for Com- putational Linguistics. · DOI · Paper page
- George Michalopoulos, Kyle Williams, Gagandeep Singh, and Thomas Lin. 2022. MedicalSum: A guided clinical abstractive summarization model for generating medical reports from patient-doctor con- versations. In Findings of the Association for Com- putational Linguistics: EMNLP 2022, pages 4741- 4749, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics. · DOI
- OpenAI. 2023. Gpt-4 technical report. arXiv preprint, arXiv/2303.08774.
- Jason Phang, Yao Zhao, and Peter Liu. 2023. Investigat- ing efficiently extending transformers for long input summarization. In Proceedings of the 2023 Con- ference on Empirical Methods in Natural Language Processing, pages 3946-3961, Singapore. Associa- tion for Computational Linguistics. · DOI
- Thomas Schaaf, Longxiang Zhang, Alireza Bayeste- htashk, Mark Fuhs, Shahid Durrani, Susanne Burger, Monika Woszczyna, and Thomas Polzin. 2021. Are you dictating to me? detecting embedded dictations in doctor-patient conversations. In 2021 IEEE Auto- matic Speech Recognition and Understanding Work- shop (ASRU), pages 587-593. · DOI
- Gagandeep Singh, Yue Pan, Jesus Andres-Ferrer, Miguel Del-Agua, Frank Diehl, Joel Pinto, and Paul Vozila. 2023. Large scale sequence-to-sequence mod- els for clinical note generation from patient-doctor conversations. In Proceedings of the 5th Clinical Natural Language Processing Workshop, pages 138- 143, Toronto, Canada. Association for Computational Linguistics. · DOI
- Jing Su, Longxiang Zhang, Hamid Reza Hassanzadeh, and Thomas Schaaf. 2022. Extract and Abstract with BART for Clinical Notes from Doctor-Patient Con- versations. In Proc. Interspeech 2022, pages 2488- 2492. · DOI
- Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in Neural Information Pro- cessing Systems, volume 30.
- Wen-Wai Yim, Yujuan Fu, Asma Ben Abacha, Neal Snider, Thomas Lin, and Meliha Yetisgen. 2023. Aci- bench: a novel ambient clinical intelligence dataset for benchmarking automatic visit note generation. Scientific Data, 10. · DOI · Paper page
- Zhiyuan Zeng, Jiatong Yu, Tianyu Gao, Yu Meng, Tanya Goyal, and Danqi Chen. 2024. Evaluating large lan- guage models at evaluating instruction following. In The Twelfth International Conference on Learning Representations.
- Jingqing Zhang, Yao Zhao, Mohammad Saleh, and Peter Liu. 2020. PEGASUS: Pre-training with extracted gap-sentences for abstractive summarization. In Pro- ceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Research, pages 11328-11339.
- PMLR.
- Longxiang Zhang, Renato Negrinho, Arindam Ghosh, Vasudevan Jagannathan, Hamid Reza Hassanzadeh, Thomas Schaaf, and Matthew R. Gormley. 2021. Leveraging pretrained models for automatic summa- rization of doctor-patient conversations. In Find- ings of the Association for Computational Linguis- tics: EMNLP 2021, pages 3693-3712, Punta Cana, Dominican Republic. Association for Computational Linguistics. · DOI
- Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric Xing, Hao Zhang, Joseph E Gonzalez, and Ion Stoica. 2023. Judging llm-as-a-judge with mt-bench and chatbot arena. In Advances in Neural Information Processing Systems, volume 36, pages 46595-46623.
- Daniel M. Ziegler, Nisan Stiennon, Jeffrey Wu, Tom B. Brown, Alec Radford, Dario Amodei, Paul Chris- tiano, and Geoffrey Irving. 2020. Fine-tuning lan- guage models from human preferences. arXiv preprint, arXiv/1909.08593.
Cited by
Sorted by citations
Gu, J., Jiang, X., Shi, Z., Tan, H., Zhai, X., Xu, C., Li, W., Shen, Y., Ma, S., Liu, H., et al. (2024) · arXiv · doi:10.48550/arxiv.2411.15594
- Authors
- Brake, N., & Schaaf, T.
- Venue
- Findings of the Association for Computational Linguistics: NAACL 2024
- Published
- 2024
- Pages
- 352-363
- OpenAlex
- W4401043072
- Cited by
- 6
- References
- 0
- Citations by year
- 2026: 3 · 2025: 3
- FWCI
- 1.73
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Findings of the Association for Computational Linguistics: NAACL (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Keywords
- clinical note generation, SOAP notes, large language models, LLM-based evaluation, Llama 2, inter-rater reliability, medical terminology, conditional generation, human evaluation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
