Schistosomiasis, a serious parasitic disease caused by trematode worms of the genus. caused Schistosoma, is not only harmful to health, but also slows down the development of the economy and society in the endemic areas. According to the World Health Organization (WHO), the disease is spread in 78 tropical and subtropical countries around the world and affects the lives of more than 700 million people in endemic areas, of whom at least 240 million are actually infected. Schistosomiasis therefore remains a major public health risk in low and middle income countries.

Control the disease

The transmission of schistosomiasis is due to the presence of water snails. Schistosomiasis japonica, once endemic to the Yangtze River, China, is closely related to the spread of Oncomelania hupensis Slug. After seven decades of continuous efforts based primarily on snail control and treatment with the drug praziquantel, transmission interruption was achieved in nine of the 12 previously endemic provinces, the lowest prevalence ever achieved.

China is now making great efforts to eradicate schistosomiasis by closely monitoring slug control. A nationwide survey of snails is carried out every spring and autumn and is the most important measure to combat schistosomiasis. The national snail survey is both labor-intensive and financially burdensome.

Effective spatial reassessment

The purpose of the snail survey is to locate the survival species of snails and to make predictions about areas of high transmission of schistosomiasis. Due to the large volume of these snail studies and the information from pilot sites, the enormous snail database could be used to predict the probability of survival of snails in the Yangtze River basin.

We started with a simple idea: does more survey locations mean more accurate predictions of the likelihood of snails?

We hypothesized that the original snail distribution data could be used through a resample approach to obtain new distribution points. Finally, the prediction performance could be compared to choose the best resampling data.

In this study, the spatial reassessment process was first developed, which defines an ecological grid cell. These cells share the same environments for snail survival. If there are two or more snail locations in this grid cell, we have only recorded one point.

Through this spatial reassessment process, we could get a new sample of snail data and then make model predictions for areas where schistosomiasis is transmitted.

In 2018, a total of 2369 locations in the Yangtze River basin were sampled, of which 1061 contained live snails (detection rate is 0.448). We set the grid cell spacing as 5 km, 10 km, 50 km, 100 km and 150 km. The re-sampling of snail sites was 1747, 1421, 209, 98 and 44, respectively. The snail detection rate is 0.462, 0.471, 0.449, 0.469 and 0.477, respectively.

As the size of the grid cells increased, the ecological zone got larger and at the same time the snail resampling locations decreased, but the snail detection rate remains stable as shown in the picture below.

Re-sampling of the snail locations according to ecological zones with different grids. From https://idpjournal.biomedcentral.com/articles/10.1186/s40249-021-00852-1

Machine learning application

After resampling the snail datasets, we attempted to combine the data with environmental and ecological variables, altitude, water distance, temperature, and rainfall to predict the suitability of snail survival areas.

The predictions are based on a machine-learning random forest algorithm. which could make more accurate predictions and also compare different spatial reassessments of the screw performance.

We have found that setting the grid cell spacing to 5 km will give the model performance the same results as the original auger survey location (see figure below).

Predicting the survival areas of snails through machine learning. From https://idpjournal.biomedcentral.com/articles/10.1186/s40249-021-00852-1

Progress for the future

Progress has been made in establishing a monitoring tool for routine work in schistosomiasis. By defining a 5 km long ecological snail surveying zone, the same results could be achieved as with the previous forecast, in which the snail surveying locations were reduced from 2369 to 1747 points. Machine learning is a useful modeling technique to predict the risk of transmission of schistosomiasis. The benefits of machine learning and spatial resampling are to ease the research effort and improve the public health disease surveillance system.

LEAVE A REPLY

Please enter your comment!
Please enter your name here