Geographical and meteorological factors have been reported to influence the prevalence of echinococcosis, but there's a lack of indicator system and model. To provide further insight into the impact of geographical and meteorological factors on AE prevalence and establish a theoretical basis for prevention and control. Principal component and regression analysis were used to screen and establish a three-level indicator system. Relative weights were examined to determine the impact of each indicator, and five mathematical models were compared to identify the best predictive model for AE epidemic levels. By analyzing the data downloaded from the China Meteorological Data Service Center and Geospatial Data Cloud, we established the KCBIS, including 50 basic indicators which could be directly obtained online, 15 characteristic indicators which were linear combination of the basic indicators and showed a linear relationship with AE epidemic, and 8 key indicators which were characteristic indicators with a clearer relationships and fewer mixed effects. The relative weight analysis revealed that monthly precipitation, monthly cold days, the difference between negative and positive temperature anomalies, basic air temperature conditions, altitude, the difference between positive and negative atmospheric pressure anomalies, monthy extremely hot days, and monthly fresh breeze days were correlated with the natural logarithm of AE prevalence, with sequential decreases in their relative weights. The multinomial logistic regression model was the best predictor at epidemic levels 1, 3, 5, and 6, whereas the CART model was the best predictor at epidemic levels 2, 4, and 5.
Read full abstract