Comparative evaluation of neural networks and tree-based models for predicting the degree of physical disability in leprosy patients

• Pedro Henrique Correia Bezerra Silva UPE
• Elisson da Silva Rocha UPE
• Patricia Takako Endo UPE
• Eraylson Galdino da Silva UPE

DOI: https://doi.org/10.5753/sbcas.2026.21602

Abstract

Leprosy is a significant public health concern due to its disabling potential and its substantial presence in Brazil. This study compared AI models applied to tabular data from the SINAN database to predict the final GIF in patients. Tree-based models (RF, LightGBM, and CatBoost) and neural networks (MLP, ResNet, and Transformer) were evaluated. LightGBM achieved superior performance and greater stability across classes, reaching an AUC OvO of 71.10%. The neural networks demonstrated a competitive performance, particularly the Transformer, which achieved an AUC OvO of 70.69%. To conclude, given the dataset used, tree-based models are more suitable for predicting GIF prognosis, while neural networks are alternatives for multimodal contexts.

References

Andrade, H. G. V. d., Rocha, E. d. S., Monteiro, K. H. d. C., de Morais, C. M., dos Santos, D. C. M., Nascimento, D. C., Dourado, R. A., Lynn, T., and Endo, P. T. (2025). On the usage of artificial intelligence in leprosy care: A systematic literature review. PLOS Computational Biology, 21(6):e1012550.

Barbieri, R. R., Xu, Y., Setian, L., Souza-Santos, P. T., Trivedi, A., Cristofono, J., Bhering, R., White, K., Sales, A. M., Miller, G., et al. (2022). Reimagining leprosy elimination with ai analysis of a combination of skin lesion images with demographic and clinical data. The Lancet Regional Health–Americas, 9.

Bomtempo, C. F., Ferrari, S. M. F., de Faria Grossi, M. A., and Lyon, S. (2023). Evolução do grau de incapacidade física e do escore olhos, mãos e pés em casos novos de hanseníase: do diagnóstico à alta medicamentosa. Hansenologia Internationalis: hanseníase e outras doenças infecciosas, 48:e37331–e37331.

Borisov, V., Leemann, T., Seßler, K., Haug, J., Pawelczyk, M., and Kasneci, G. (2021). Deep neural networks and tabular data: A survey. arXiv preprint arXiv:2110.01889.

Breiman, L. (2001). Random forests. Machine Learning, 45(1):5–32.

da Costa, N. M. G. B., Barbosa, T. d. C. S., Queiroz, D. T., Oliveira, A. K. A., Montemezzo, L. C. D., and do Couto Andrade, U. (2020). Perfil sociodemográfico e grau de incapacidade do portador de hanseníase em um centro de referência no estado do ceará. Brazilian Journal of Development, 6(6):41439–41449.

da Silva, Y. E. D., Salgado, C. G., Conde, V. M. G., and Conde, G. A. B. (2018). Application of clustering technique with kohonen self-organizing maps for the epidemiological analysis of leprosy. In Advances in Intelligent Systems and Computing, pages 295–309. Springer.

De Souza, M. L. M., Lopes, G. A., Branco, A. C., Fairley, J. K., and Fraga, L. A. D. O. (2021). Leprosy screening based on artificial intelligence: Development of a crossplatform app. JMIR MHealth and UHealth, 9(4):e23718.

DKE (2022). German Standardization Roadmap on Artificial Intelligence. DIN e. V., Berlin, 2 edition.

Dorogush, A. V., Ershov, V., and Gulin, A. (2018). Catboost: Unbiased boosting with categorical features. In Proceedings of the 32nd International Conference on Machine Learning (ICML) – Workshop on Challenges for Categorical Data. PMLR.

Dutra da Silva, Y. E., Salgado, C. G., Gomes Conde, V. M., and Barros Conde, G. A. (2018). Data mining using clustering techniques as leprosy epidemiology analyzing model. In Lecture Notes in Computer Science, pages 284–293. Springer.

Fernandes, J. R. N., Teles, A. S., Fernandes, T. R. S., Lima, L. D. B., Balhara, S., Gupta, N., and Teixeira, S. (2024). Artificial intelligence on diagnostic aid of leprosy: A systematic literature review. Journal of Clinical Medicine, 13(180):1–22.

Freitas, L. R. S., Freitas, J. A. O., Penna, G. O., and Duarte, E. C. (2025). Evaluating machine learning models for predicting late leprosy diagnosis by physical disability grade in brazil (2018–2022). Tropical Medicine and Infectious Disease, 10(5):131.

Gorishniy, Y., Rubachev, I., Khrulkov, V., and Babenko, A. (2021). Revisiting deep learning models for tabular data. In Advances in Neural Information Processing Systems (NeurIPS).

Grinsztajn, L., Oyallon, E., and Varoquaux, G. (2022). Why do tree-based models still outperform deep learning on typical tabular data? In Advances in Neural Information Processing Systems (NeurIPS).

Gutta, V., Ganakammal, S. R., Jones, S., Beyers, M., and Chandrasekaran, S. (2024). Unnt: A novel utility for comparing neural net and tree-based models. PLOS Computational Biology, 20(4):1–11.

Kadra, A., Lindauer, M., Hutter, F., and Grabocka, J. (2021). Well-tuned simple nets excel on tabular datasets. NIPS ’21, Red Hook, NY, USA. Curran Associates Inc.

Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q., and Liu, T.-Y. (2017). Lightgbm: A highly efficient gradient boosting decision tree. In Proceedings of the 31st Conference on Neural Information Processing Systems (NeurIPS), Long Beach, CA, USA.

Ministério da Saúde (2022). Protocolo clínico e diretrizes terapêuticas da hanseniase. Publicação oficial; Accessed 2025-12-09.

Organização Mundial da Saúde (2010). Estratégia Global Aprimorada para Redução Adicional da Carga da Hanseníase : 2011–2015 — Diretrizes Operacionais (Atualizadas). Organização Pan-Americana da Saúde, Brasília.

Salmi, M., Atif, D., Oliva, D., Abraham, A., and Ventura, S. (2024). Handling imbalanced medical datasets: review of a decade of research. Artificial Intelligence Review, 57(10):273.

Saltz, J. S. (2021). Crisp-dm for data science: strengths, weaknesses and potential next steps. In 2021 IEEE International Conference on Big Data (Big Data), pages 2337–2344. IEEE.

Santos, G. M. C. d., Byrne, R. L., Cubas-Atienzar, A. I., and Santos, V. S. (2024). Factors associated with delayed diagnosis of leprosy in an endemic area in northeastern brazil: a cross-sectional study. Cadernos de Saúde Pública, 40(1):e00113123.

Secretaria de Vigilância em Saúde (2023). Formulário para avaliação neurológica simplificada e classificação do grau de incapacidade física em hanseníase.

Shmuel, A., Glickman, O., and Lazebnik, T. (2025). A comprehensive benchmark of machine and deep learning models on structured data for regression and classification. Neurocomputing, 655:131337. Available online 3 September 2025.

Shwartz-Ziv, R. and Armon, A. (2022). Tabular data: Deep learning is not all you need. Information Fusion.

Véras, G. C. B., da Silva, L. H., Sarmento, W. M., de Moraes, R. M., dos Santos Oliveira, S. H., and Soares, M. J. G. O. (2023). Características sociodemográficas e epidemiológicas relacionadas ao grau de incapacidade física em hanseníase no estado da paraíba, brasil. Hansenologia Internationalis: hanseníase e outras doenças infecciosas, 48:e38999–e38999.

Véras, G. C. B., Lima Júnior, J. F., Cândido, E. L., and Maia, E. R. (2021). Risk factors for physical disability due to leprosy: a case-control study. Cadernos Saúde Coletiva, 29:411–423.

Wirth, R. and Hipp, J. (2000). Crisp-dm: Towards a standard process model for data mining. In Proceedings of the 4th international conference on the practical applications of knowledge discovery and data mining, volume 1, pages 29–39. Manchester.

World Health Organization (2020). Ending the Neglect to Attain the Sustainable Development Goals: A Road Map for Neglected Tropical Diseases 2021–2030. World Health Organization, Geneva. Electronic version. Licensed under CC BY-NC-SA 3.0 IGO.

World Health Organization (2021). Global leprosy (hansen’s disease) strategy 2021–2030: Towards zero leprosy. Technical report, World Health Organization, Geneva. Global strategy document.

World Health Organization (2025). Global leprosy (hansen disease) update, 2024: Beyond zero cases – what elimination of leprosy really means. Weekly Epidemiological Record, 100(37):365–384. Published 12 September 2025.