Ecological Epidemiology Insights into Clonorchiosis Endemicity in Southern China and Northern Vietnam: A Comprehensive Machine Learning Analysis

Abstract

Background: Clonorchis sinensis, the liver fluke responsible for clonorchiosis, poses significant public health challenges in Southern China and Northern Vietnam. Understanding its transmission dynamics is crucial for effective public health interventions. Objectives: This study employs an ecological epidemiological approach, integrating environmental, climatic, and socio-cultural factors, to model and predict the transmission patterns of C. sinensis in these regions. Methods: Leveraging machine learning techniques, we analyzed data from systematic literature reviews and national health surveys conducted between 2000 and 2018. Environmental factors, climate variables, and socio-cultural practices, notably raw fish consumption, were examined to assess their impact on the endemic of C. sinensis. Results: Our analysis identifies raw fish consumption as a crucial determinant of C. sinensis transmission. The study revealed that 54.9% of counties in Guangxi Province and 31.7% of provincial-level divisions in Vietnam documented raw fish consumption, correlating with higher infection probabilities. Notably, logistic regression models achieved an area under the curve (AUC) of 0.941, demonstrating high predictive accuracy. Environmental comparisons showed significant differences between two places, with Vietnam showed a higher annual mean temperature (Bio1: 23.37°C vs. 20.86°C), greater temperature seasonality (Bio4: 464.92 vs. 609.33), and more annual precipitation (Bio12: 1731.64mm vs. 1607.56mm) compared to Guangxi, all factors contributing to varying levels of endemicity. These spatial analyses identified key high-risk areas, particularly along the China-Vietnam border, highlighting zones requiring targeted public health interventions. Conclusion: This study underscores the interplay of ecological and socio-cultural factors in the transmission of clonorchiosis. The predictive models developed offer valuable insights for public health strategies, emphasizing the need for regional cooperation in disease control and prevention. Our approach demonstrates the potential of integrating diverse data sources in ecological epidemiology to address complex public health challenges.