Text Style Transfer with Large Language Models: Enhancing Medical Anamnesis Transcriptions

• Yanna Torres Gonçalves Universidade Federal do Ceará (UFC)
• Ticiana L. Coelho da Silva Universidade Federal do Ceará (UFC)

DOI: https://doi.org/10.5753/sbbd_estendido.2025.247598

Resumo

This study investigates the use of Large Language Models (LLMs) for enhancing medical anamnesis transcriptions via Text Style Transfer (TST). It involves benchmarking three models (Phi3, Llama, and Mistral) and fine-tuning Mistral based on evaluator feedback. Mistral achieved the best initial performance but showed limited improvement after fine-tuning. The work highlights the challenges of adapting LLMs to clinical tasks and discusses limitations such as data quality and evaluator bias. Future directions include extended training, dataset expansion, and exploring new machine learning techniques.

Palavras-chave: Text Style Transfer, Large Language Models, Medical History

Referências

Fleiss, J. L. (1971). Measuring nominal scale agreement among many raters. Psychological bulletin, 76(5):378.

Gonçalves, Y., Alves, J., Sá, B., Silva, L., Macedo, J., and da Silva, T. C. (2024). Speech recognition models in assisting medical history. In Anais do XXXIX Simpósio Brasileiro de Bancos de Dados, pages 485–497, Porto Alegre, RS, Brasil. SBC.

Gür, B. (2012). Improving speech recognition accuracy for clinical conversations. PhD thesis, Massachusetts Institute of Technology.

Jin, D., Jin, Z., Hu, Z., Vechtomova, O., and Mihalcea, R. (2022). Deep learning for text style transfer: A survey. Computational Linguistics, 48(1):155–205.

Kendall, M. G. (1938). A new measure of rank correlation. Biometrika, 30(1-2):81–93.

Lai, W., Hangya, V., and Fraser, A. (2024). Style-specific neurons for steering llms in text style transfer. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 13427–13443.

Lee, J., Yoon, W., Kim, S., Kim, D., Kim, S., So, C. H., and Kang, J. (2019). BioBERT: a pre-trained biomedical language representation model for biomedical text mining. Bioinformatics, 36(4):1234–1240.

Liu, Q., Qin, J., Ye, W., Mou, H., He, Y., and Wang, K. (2024). Adaptive prompt routing for arbitrary text style transfer with pre-trained language models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 18689–18697.

Luo, R., Sun, L., Xia, Y., Qin, T., Zhang, S., Poon, H., and Liu, T.-Y. (2022). BioGPT: generative pre-trained transformer for biomedical text generation and mining. Briefings in Bioinformatics, 23(6):bbac409.

Mukherjee, S. and Dušek, O. (2024). Text style transfer: An introductory overview.

Mukherjee, S., Ojha, A. K., and Dušek, O. (2024). Are large language models actually good at text style transfer? arXiv preprint arXiv:2406.05885.

Papineni, K., Roukos, S., Ward, T., and Zhu, W.-J. (2002). Bleu: A method for automatic evaluation of machine translation. In Proceedings of the 40th ACL, page 311–318, USA. Association for Computational Linguistics.

Reddy, D. (1976). Speech recognition by machine: A review. Proceedings of the IEEE, 64(4):501–531.

Soares, M. O. M., Higa, E. d. F. R., Gomes, L. F., Marvã, J. P. Q., da Fonseca Gomes, A. I., and Gonçalves, A. H. C. (2016). Impacto da anamnese para o cuidado integral: visão dos estudantes portugueses. Revista Brasileira em Promoção da Saúde, 29:66–75.

Yehia, A. C., Viana, P. R. L., Macedo, M. V. M., de Souza Dias, N. C., Campos, C. C., Jardim, S. N., and de Almeida Garcia, J. N. A. (2024). Anamnese na prática clínica: uma revisão sobre suas aplicações e importância. Revista da Sociedade Brasileira de Clínica Médica, 22(2):116–120.