Japiim Health: Democratizing Access to Brazilian Arbovirus Data via Retrieval-Augmented Generation on WhatsApp

• Ana Rita Marega Gonçalves Universidade de São Paulo (USP)
• Nicolas de Souza Maia Universidade de São Paulo (USP)
• Melissa de Souza Maia Universidade de São Paulo (USP)
• Marina Carvalho Arruda Barreto Universidade de São Paulo (USP)
• Robson Parmezan Bonidia Universidade Tecnológica Federal do Paraná (UTFPR)
• André de Carvalho Universidade de São Paulo (USP)

DOI: https://doi.org/10.5753/sbbd_estendido.2026.249582

Resumo

Although the “Sistema de Informação de Agravos de Notificação” is fundamental for epidemiological surveillance in Brazil, its complex tabular format and the lack of contextual interpretation in existing visualization tools create significant barriers to data accessibility. In this context, we present Japiim, a multimodal chatbot integrated with WhatsApp, developed to democratize access to national microdata on arboviruses. Based on a Retrieval-Augmented Generation (RAG) architecture and the Sabiá-4 language model, Japiim processes granular health records, transforming them into intuitive texts, voice responses, and dynamic charts. A comparative analysis with general-purpose language models (ChatGPT, Gemini, Claude) reveals that Japiim mitigates quantitative hallucinations and imposes strict safety limits, avoiding the generation of inappropriate medical advice. By combining a widely adopted and low-friction communication platform with a foundation in reliable data, Japiim empowers health managers and citizens with precise and immediate information on public health.

Palavras-chave: Retrieval-Augmented Generation (RAG), Large Language Models (LLMs), Epidemiological surveillance, WhatsApp Chatbot, Data democratization, Public health, SINAN, Arboviruses

Referências

Barkhad, A., Santos, G., R C Campos, S., M A Braz, L., P Freitas, L., Zinszer, K., de Souza, R., Waldron, I., Loeb, M., Luna, E., and Mbuagbaw, L. (2026). Knowledge, attitudes, practices and perceptions of the eco-bio-social determinants of dengue transmission in são paulo, brazil: A mixed-methods study. Trop Med Internation Health, 31(1):58–79.

Chiat, J. et al. (2024). Artificial intelligence, chatgpt, and other large language models for social determinants of health: Current state and future directions. Cell Reports Medicine, 5(1):101356–101356.

Comini, N., Gozzi, N., and Perra, N. (2024). Bridging brazil’s digital divide: How internet inequality mirrors income gaps. World Bank Blogs. Accessed: 2026-05-18.

Es, S., James, J., Anke, L. E., and Schockaert, S. (2024). Ragas: Automated evaluation of retrieval augmented generation. In Proceedings of the 18th conference of the european chapter of the association for computational linguistics: system demonstrations, pages 150–158.

Fernandez, M., Pinto, H. A., Fernandes, L. M. M., Oliveira, J. A. S. d., Lima, A. M. F. d. S., Santana, J. S. S., and Chioro, A. (2025). Interoperability in universal healthcare systems: insights from brazil’s experience integrating primary and hospital health care data. Frontiers in Digital Health, Volume 7 - 2025.

Gurgel-Gonçalves, R., Oliveira, W. K. d., and Croda, J. (2024). The greatest dengue epidemic in brazil: Surveillance, prevention, and control. Rev Soc Bras Med Trop, 57.

Henriques, C. M. P., Moura, N. F. O., and Souza, P. B. (2024). Challenges and lessons from the covid-19 pandemic for health surveillance in brazil: reflections on technologies, models, and system organization. Rev Bras Epidemiol, 27:e240049.

Ministério da Saúde (2025). Portaria gm/ms no 6.734, de 18 de março de 2025. Disponível em: [link].

Qin, H. and Tong, Y. (2025). Opportunities and challenges for large language models in primary health care. J Primar Care & Commun Health, 16.

Wu, E., Balise, R., Katz, B., Harris, D., Bullard, M., Fareed, N., Larochelle, M., and Villani, J. (2026). Building public health data dashboards: Tutorial playbook. JMIR public health and surveillance, 12:e83157.