Abstract：
Our main aim was to estimate and compare the effects of six environmental variables (air temperature, soil temperature, rainfall, runoff, soil moisture, and the enhanced vegetation index) on excess cases of cutaneous leishmaniasis in Colombia. We used epidemiological data from the Colombian Public Health Surveillance System (January 2007 to December 2019). Environmental data were obtained from remote sensing sources including the National Oceanic and Atmospheric Administration, the Global Land Data Assimilation System (GLDAS), and the Moderate Resolution Imaging Spectroradiometer. Data on population were obtained from the TerriData dataset. We implemented a causal inference approach using a machine learning algorithm to estimate the causal association of the environmental variables on the monthly occurrence of excess cases of cutaneous leishmaniasis. The results showed that the largest causal association corresponded to soil moisture with a lag of 3 months, with an average increase of 8.0% (95% confidence interval [CI] 7.7-8.3%) in the occurrence of excess cases. The temperature-related variables (air temperature and soil temperature) had a positive causal effect on the excess cases of cutaneous leishmaniasis. It is noteworthy that rainfall did not have a statistically significant causal effect. This information could potentially help to monitor and control cutaneous leishmaniasis in Colombia, providing estimates of causal effects using remote sensor variables.
摘要：
我们的主要目的是估计和比较六个环境变量(气温，土壤温度,降雨,径流,土壤湿度,和增强的植被指数）对哥伦比亚皮肤利什曼病的过量病例。我们使用哥伦比亚公共卫生监测系统（2007年1月至2019年12月）的流行病学数据。环境数据来自遥感来源，包括国家海洋和大气管理局，全球土地数据同化系统(GLDAS)，和中分辨率成像光谱辐射计。人口数据来自TerriData数据集。我们使用机器学习算法实施了因果推理方法，以估计环境变量对每月发生的皮肤利什曼病病例的因果关系。结果表明，最大的因果关联对应于土壤水分，滞后3个月，多余病例的发生率平均增加8.0％（95％置信区间[CI]7.7-8.3％）。与温度相关的变量（气温和土壤温度）对皮肤利什曼病的过量病例具有积极的因果关系。值得注意的是，降雨没有统计上显着的因果关系。这些信息可能有助于监测和控制哥伦比亚的皮肤利什曼病，使用遥感器变量提供因果效应的估计。
