Predição de Surtos de Dengue e Diagnóstico de Sífilis Congênita Utilizando Aprendizado de Máquina
Abstract
Congenital syphilis and dengue are two diseases that cause significant impacts in Brazil and other countries in the Southern Hemisphere, affecting the health of millions of people. Syphilis is a sexually transmitted infection (STI) that, when transmitted to children during pregnancy, is called congenital syphilis. Dengue is a viral disease transmitted by the Aedes Aegypti and Aedes Albopictus mosquitoes. In this thesis, we developed innovative applications of machine learning models for these diseases. The first one estimates the probability of a child being born with syphilis. The second predicts dengue outbreaks based on sociodemographic and climatic data, historical series of cases, number of health units, mosquito measurement index, and historical series of zika and chikungunya. In the case of congenital syphilis, we evaluated the models using the AUC (Area Under Curve) metric and the result was good but not excellent, i.e., 0.68 for the prediction of positive cases, obtained by the LightGBM and XGBoost models. With regard to dengue, the Catboost model obtained very good results, identifying 75% of outbreaks three months in advance. A significant part of this work was invested in the explanation of dengue predictions, which makes the model an important ally for the design of public health policies.
References
Guinsburg, R. and Santos, A. M. N. d. (2010). Critérios diagnósticos e tratamento da sífilis congênita.
Hyndman, R. J. and Athanasopoulos, G. (2018). Forecasting: principles and practice. OTexts.
Lago, E. G., Rodrigues, L. C., Fiori, R. M., and Stein, A. T. (2004). Congenital syphilis: identification of two distinct profiles of maternal characteristics associated with risk. Sexually transmitted diseases, 31(1):33–37.
Liu, J.-B., Hong, F.-C., Pan, P., Zhou, H., Yang, F., Cai, Y.-M., Wen, L.-Z., Lai, Y.-H., Lin, L.-J., and Zeegers, M. P. (2010). A risk model for congenital syphilis in infants born to mothers with syphilis treated in gestation: a prospective cohort study. Sexually Transmitted Infections, 86(4):292–296.
Lundberg, S. M., Erion, G., Chen, H., DeGrave, A., Prutkin, J. M., Nair, B., Katz, R., Himmelfarb, J., Bansal, N., and Lee, S.-I. (2020). From local explanations to global understanding with explainable ai for trees. Nature machine intelligence, 2(1):56–67.
Siriyasatien, P., Chadsuthi, S., Jampachaisri, K., and Kesorn, K. (2018). Dengue epidemics prediction: A survey of the state-of-the-art based on data science processes. IEEE Access, 6:53757–53795.
