UNASSIGNED: Historical records of sand fly vectors, human leishmaniasis, bioclimatic indicators, and environmental variables were integrated in a machine learning framework (XGBoost) to predict suitability in two past periods (2001-2010 and 2011-2020). We further assessed if predictions were associated with human and animal disease data from selected countries (France, Greece, Italy, Portugal, and Spain).
UNASSIGNED: An increase in the number of climatically suitable regions for leishmaniasis was detected, especially in southern and eastern countries, coupled with a northward expansion towards central Europe. The final model had excellent predictive ability (AUC = 0.970 [0.947-0.993]), and the suitability predictions were positively associated with human leishmaniasis incidence and canine seroprevalence for Leishmania.
UNASSIGNED: This study demonstrates how key epidemiological data can be combined with open-source climatic and environmental information to develop an indicator that effectively tracks spatiotemporal changes in climatic suitability and disease risk. The positive association between the model predictions and human disease incidence demonstrates that this indicator could help target leishmaniasis surveillance to transmission hotspots.
UNASSIGNED: European Union Horizon Europe Research and Innovation Programme (European Climate-Health Cluster), United Kingdom Research and Innovation.
■沙蝇媒介的历史记录,人类利什曼病,生物气候指标,和环境变量被集成在机器学习框架(XGBoost)中,以预测过去两个时期(2001-2010年和2011-2020年)的适用性。我们进一步评估了预测是否与选定国家的人类和动物疾病数据相关(法国,希腊,意大利,葡萄牙,和西班牙)。
■检测到适合利什曼病的气候区域数量增加,特别是在南部和东部国家,再加上向中欧向北扩张。最终模型具有出色的预测能力(AUC=0.970[0.947-0.993]),适合性预测与利什曼原虫的人类利什曼病发病率和犬血清阳性率呈正相关。
■这项研究展示了如何将关键的流行病学数据与开源的气候和环境信息相结合,以开发出一种指标,该指标可以有效地跟踪气候适宜性和疾病风险的时空变化。模型预测与人类疾病发病率之间的正相关表明,该指标可以帮助将利什曼病监测目标定位到传播热点。
■欧盟地平线欧洲研究与创新计划(欧洲气候-健康集群),英国研究与创新。