Inteligência Artificial Generativa com Dados Limitados: Superando Desafios na Avaliação Automática da Dor Neonatal
Resumo
Esta trabalho investiga o potencial da Inteligência Artificial Generativa (IAGen) na neonatologia, focando na superação da escassez de dados para a avaliação automática da dor neonatal. A pesquisa resultou na criação do SynUnifesp, um conjunto de dados composto inteiramente por imagens sintéticas de faces de recém-nascidos, utilizado para expandir bases de dados limitadas. Para aprimorar a avaliação desses modelos generativos, foi proposta a métrica Fréchet AutoEncoder Distance (FAED), que utiliza o espaço latente para garantir maior robustez contra perturbações e melhor generalização que o FID tradicional. A eficácia do SynUnifesp foi validada em tarefas de classificação sob três paradigmas de Inteligência Artificial: aprendizado supervisionado, de poucas amostras (few-shot) e auto-supervisionado. Os resultados demonstram que a inclusão de dados sintéticos proporciona ganhos significativos no desempenho da avaliação automática da dor neonatal, consolidando a IAGen como uma ferramenta viável para avanços clínicos em Unidades de Terapia Intensiva Neonatal.Referências
Beam, K., Sharma, P., Levy, P., and Beam, A. L. (2024). Artificial intelligence in the neonatal intensive care unit: the time is now. Journal of Perinatology, 44(1):131–135.
Borji, A. (2022). Pros and cons of gan evaluation measures: New developments. Computer Vision and Image Understanding, 215:103329.
Buzuti, L., Giraldi, G., Heiderich, T., Barros, M., Guinsburg, R., and Thomaz, C. (2024). Generative ai for neonatal pain assessment: a sound approach to improve data-driven automatic recognition in intensive care unit. Available at SSRN 4994076.
Buzuti, L. F. and Thomaz, C. E. (2023). Fréchet autoencoder distance: A new approach for evaluation of generative adversarial networks. Computer Vision and Image Understanding, page 103768.
Deng, J., Guo, J., Ververas, E., Kotsia, I., and Zafeiriou, S. (2020). Retinaface: Single-shot multi-level face localisation in the wild. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).
He, K., Zhang, X., Ren, S., and Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778.
Heiderich, T. M., Leslie, A. T. F. S., and Guinsburg, R. (2015). Neonatal procedural pain can be assessed by computer software that has good sensitivity and specificity to detect facial movements. Acta Paediatrica, 104(2):e63–e69.
Hinton, G., Deng, L., Yu, D., Dahl, G. E., Mohamed, A.-r., Jaitly, N., Senior, A., Vanhoucke, V., Nguyen, P., Sainath, T. N., et al. (2012). Deep neural networks for acoustic modeling in speech recognition: The shared views of four research groups. IEEE Signal processing magazine, 29(6):82–97.
Karen, S. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv: 1409.1556.
Kingma, D. P. and Welling, M. (2013). Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114.
Klushyn, A., Chen, N., Kurle, R., Cseke, B., and van der Smagt, P. (2019). Learning hierarchical priors in vaes. Advances in neural information processing systems, 32.
Oquab, M., Darcet, T., Moutakanni, T., Vo, H. V., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., Howes, R., Huang, P.-Y., Xu, H., Sharma, V., Li, S.-W., Galuba, W., Rabbat, M., Assran, M., Ballas, N., Synnaeve, G., Misra, I., Jegou, H., Mairal, J., Labatut, P., Joulin, A., and Bojanowski, P. (2023). Dinov2: Learning robust visual features without supervision.
Rve, G. (1987). Pain expression in neonates: facial action and cry. Pain, 28:395–410.
Snell, J., Swersky, K., and Zemel, R. (2017). Prototypical networks for few-shot learning. Advances in neural information processing systems, 30.
Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., and Jegou, H. (2021). Training data-efficient image transformers amp; distillation through attention. In International Conference on Machine Learning, volume 139, pages 10347–10357.
Vahdat, A. and Kautz, J. (2020). Nvae: A deep hierarchical variational autoencoder. Advances in neural information processing systems, 33:19667–19679.
Vinyals, O., Blundell, C., Lillicrap, T., Wierstra, D., et al. (2016). Matching networks for one shot learning. Advances in neural information processing systems, 29.
Zamzmi, G., Paul, R., Goldgof, D., Kasturi, R., and Sun, Y. (2019). Pain assessment from facial expression: Neonatal convolutional neural network (n-cnn). In 2019 International Joint Conference on Neural Networks (IJCNN), pages 1–7. IEEE.
Borji, A. (2022). Pros and cons of gan evaluation measures: New developments. Computer Vision and Image Understanding, 215:103329.
Buzuti, L., Giraldi, G., Heiderich, T., Barros, M., Guinsburg, R., and Thomaz, C. (2024). Generative ai for neonatal pain assessment: a sound approach to improve data-driven automatic recognition in intensive care unit. Available at SSRN 4994076.
Buzuti, L. F. and Thomaz, C. E. (2023). Fréchet autoencoder distance: A new approach for evaluation of generative adversarial networks. Computer Vision and Image Understanding, page 103768.
Deng, J., Guo, J., Ververas, E., Kotsia, I., and Zafeiriou, S. (2020). Retinaface: Single-shot multi-level face localisation in the wild. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).
He, K., Zhang, X., Ren, S., and Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778.
Heiderich, T. M., Leslie, A. T. F. S., and Guinsburg, R. (2015). Neonatal procedural pain can be assessed by computer software that has good sensitivity and specificity to detect facial movements. Acta Paediatrica, 104(2):e63–e69.
Hinton, G., Deng, L., Yu, D., Dahl, G. E., Mohamed, A.-r., Jaitly, N., Senior, A., Vanhoucke, V., Nguyen, P., Sainath, T. N., et al. (2012). Deep neural networks for acoustic modeling in speech recognition: The shared views of four research groups. IEEE Signal processing magazine, 29(6):82–97.
Karen, S. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv: 1409.1556.
Kingma, D. P. and Welling, M. (2013). Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114.
Klushyn, A., Chen, N., Kurle, R., Cseke, B., and van der Smagt, P. (2019). Learning hierarchical priors in vaes. Advances in neural information processing systems, 32.
Oquab, M., Darcet, T., Moutakanni, T., Vo, H. V., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., Howes, R., Huang, P.-Y., Xu, H., Sharma, V., Li, S.-W., Galuba, W., Rabbat, M., Assran, M., Ballas, N., Synnaeve, G., Misra, I., Jegou, H., Mairal, J., Labatut, P., Joulin, A., and Bojanowski, P. (2023). Dinov2: Learning robust visual features without supervision.
Rve, G. (1987). Pain expression in neonates: facial action and cry. Pain, 28:395–410.
Snell, J., Swersky, K., and Zemel, R. (2017). Prototypical networks for few-shot learning. Advances in neural information processing systems, 30.
Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., and Jegou, H. (2021). Training data-efficient image transformers amp; distillation through attention. In International Conference on Machine Learning, volume 139, pages 10347–10357.
Vahdat, A. and Kautz, J. (2020). Nvae: A deep hierarchical variational autoencoder. Advances in neural information processing systems, 33:19667–19679.
Vinyals, O., Blundell, C., Lillicrap, T., Wierstra, D., et al. (2016). Matching networks for one shot learning. Advances in neural information processing systems, 29.
Zamzmi, G., Paul, R., Goldgof, D., Kasturi, R., and Sun, Y. (2019). Pain assessment from facial expression: Neonatal convolutional neural network (n-cnn). In 2019 International Joint Conference on Neural Networks (IJCNN), pages 1–7. IEEE.
Publicado
01/06/2026
Como Citar
BUZUTI, Lucas F.; THOMAZ, Carlos E..
Inteligência Artificial Generativa com Dados Limitados: Superando Desafios na Avaliação Automática da Dor Neonatal. In: PRÊMIO ARTUR ZIVIANI - CONCURSO DE TESES E DISSERTAÇÕES (DOUTORADO) - SIMPÓSIO BRASILEIRO DE COMPUTAÇÃO APLICADA À SAÚDE (SBCAS), 26. , 2026, Ouro Preto/MG.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 138-143.
ISSN 2763-8987.
DOI: https://doi.org/10.5753/sbcas_estendido.2026.20308.
