Mining Discursive and Interactional Signals in Public Teleconsultation Platforms: A Deep Learning Approach to Referral Conduct Analysis

• Diego S. N. dos Santos UFOP
• Alvaro B. Cunha UFOP
• Gustavo Penna UFMG
• Frederico Coelho UFMG
• Carlos H. G. Ferreira UFOP
• Luiz Carlos B. Torres UFOP

DOI: https://doi.org/10.5753/sbcas_estendido.2026.26520

Resumo

Digital platforms used in public healthcare systems generate large volumes of interactional and discursive data. In the Brazilian Unified Health System (SUS), teleconsultation platforms mediate communication between primary care professionals and specialists, producing textual dialogues that encode referral requests, diagnostic uncertainty, collaborative decision-making, and operational patterns of care coordination. This paper investigates whether such digital and discursive signals can be mined to classify patient conduct as Retention in primary care or Referral to specialized care. Using 40,506 deidentified teleconsultation records from a Brazilian telehealth center, we evaluate Transformer-based models adapted to Portuguese clinical text, combining BioBERTpt, Head+Tail truncation, Focal Loss, and validation-based threshold selection. The final model achieved a Macro F1-Score of 0.61 and an accuracy of 71.2%, preserving predictive performance while shifting the validationoptimized threshold closer to the default probability boundary. Beyond aggregate performance, we conduct a post-hoc analysis across medical specialties, revealing statistically significant heterogeneity in model errors. These findings suggest that interprofessional teleconsultation logs can serve as a valuable source of digital and discursive signals for operational public health analysis, while also highlighting the need for subgroup evaluation, calibration-specific metrics, and external validation before any prospective deployment.

Referências

Andrade, J., Duarte, J., Shimaoka, A., Costa, T., Bandiera-Paiva, P., Junior, A., Silva, G., and Pacheco, E. (2024). Natural language processing for triage of cerebral large-vessel occlusion. Arquivos de Neuro-Psiquiatria, 82:1–8.

Brasil (2018). Lei nº 13.709, de 14 de agosto de 2018. Lei Geral de Proteção de Dados Pessoais (LGPD). Acesso em: 13 mai. 2026.

Cho, S., Lee, M., Yu, J., Yoon, J., Choi, J.-B., Jung, K.-H., and Cho, J. (2024). Leveraging large language models for improved understanding of communications with patients with cancer in a call center setting: Proof-of-concept study. J Med Internet Res, 26:e63892.

Cunha Reis, T. (2025). Artificial intelligence and natural language processing for improved telemedicine: Before, during and after remote consultation. Atención Primaria, 57(8):103228.

Feizollah, A., Lin, C.-Y., O’Malley, L., Thompson, W., Listl, S., and Byrne, M. (2025). The use of natural language processing to interpret unstructured patient feedback on health services: Scoping review. J Med Internet Res, 27:e72853.

Lin, T.-Y., Goyal, P., Girshick, R., He, K., and Dollár, P. (2020). Focal loss for dense object detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(2):318–327.

Molenaar, A., Lukose, D., Brennan, L., Jenkins, E. L., and McCaffrey, T. A. (2024). Using natural language processing to explore social media opinions on food security: Sentiment analysis and topic modeling study. J Med Internet Res, 26:e47826.

Raff, D., Stewart, K., Yang, M. C., Shang, J., Cressman, S., Tam, R., Wong, J., Tammemägi, M. C., and Ho, K. (2024). Improving triage accuracy in prehospital emergency telemedicine: Scoping review of machine learning–enhanced approaches. Interact J Med Res, 13:e56729.

Scherbakov, D. A., Hubig, N. C., Lenert, L. A., Alekseyenko, A. V., and Obeid, J. S. (2025). Natural language processing and social determinants of health in mental health research: Ai-assisted scoping review. JMIR Ment Health, 12:e67192.

Schneider, E. T. R., de Souza, J. V. A., Knafou, J., Oliveira, L. E. S. e., Copara, J., Gumiel, Y. B., Oliveira, L. F. A. d., Paraiso, E. C., Teodoro, D., and Barra, C. M. C. M. (2020). BioBERTpt - a Portuguese neural language model for clinical named entity recognition. In Rumshisky, A., Roberts, K., Bethard, S., and Naumann, T., editors, Proceedings of the 3rd Clinical Natural Language Processing Workshop, pages 65–72, Online. Association for Computational Linguistics.

Souza, F., Nogueira, R., and Lotufo, R. (2020). Bertimbau: Pretrained bert models for brazilian portuguese. In Intelligent Systems: 9th Brazilian Conference, BRACIS 2020, Rio Grande, Brazil, October 20–23, 2020, Proceedings, Part I, page 403–417, Berlin, Heidelberg. Springer-Verlag.