Esquistossomose mansônica em áreas de transmissão no Brasil: morbimortalidade, modelagem preditiva e aspectos críticos no controle da doença

Abstract

Schistosomiasis is a chronic parasitic disease caused by fluke worms of the genus Schistosoma, which affects approximately 240 million people globally. This infection occurs mainly in 78 countries in Africa, the Americas, the Middle East and Southeast Asia, where more than 700 million individuals live in endemic areas at risk of infection. Transmission is favored in underdeveloped tropical and subtropical regions, with precarious sanitary conditions and populations living in close contact with water bodies contaminated by the parasite. In the American continent, schistosomiasis is found in Central and South America, with Brazil being the country with the highest infection rates. Schistosomiasis continues to be a serious public health problem in Brazil, affecting many people who live in extensive endemic areas of the Brazilian territory. To fight schistosomiasis, Brazil adopted a public policy for disease control in Brazil and implemented the Schistosomiasis Control Program (PCE), which is based on early diagnosis and timely treatment of positive cases. In former days, it was common to find areas with high infection rates of schistosomiasis in Brazil. However, this reality has changed over the decades. This reduction in infection rates is the result of interventions implemented by the PCE, which contributed to reducing the occurrence of the disease. Currently, there is a drop in prevalence rates in endemic areas of the country, this could be encouraging, however, this reduction was also accompanied by a decrease in the number of municipalities with interventions and an active search within the PCE during the recent years. This fact may result in erroneous conclusions about the transmission profile and actual situation of the disease. For a better understanding of this process, a systematic review with meta-analysis on the prevalence of schistosomiasis in Brazil was carried out, accessing publications from the year 1980 onwards. The mean resulting prevalence from these studies was 21%, with highest concentrations in Southeastern and Northeastern Brazil, where Minas Gerais was the state with the highest number of studies. Another important fact observed about schistosomiasis in Brazil was that 43% of the possibility of evolving to death is related to the state of residence. This indicates that nearly half of the possibility of progressing to death is related to the state of residence. As demonstrated with the reduction of PCE activities, it is possible that the cases of schistosomiasis in Brazil are being underestimated. Thus, a more precise understanding and estimates on infection is important. For this, 11 machine learning algorithms were built and tested and used to estimate the occurrence of the disease. For the algorithms, the number of deaths, hospitalizations and severe cases for schistosomiasis, mesoregion and microregion of residence were used. The best model was the Random Forest Regressor, with the lowest MSE and highest R². Thus, the final model estimated a large number of undiagnosed cases, which should be approximately twice the number currently diagnosed, with the highest occurrence in the Jequitinhonha Valley, Central-South Bahia and on the coast of northeastern Brazil. Therefore, this work contributed to the understanding and the magnitude of the current situation of schistosomiasis in Brazil and proposed better strategies for surveillance and control of the disease.