Interpreting predictive maps of disease: highlighting the pitfalls of distribution models in epidemiology

Abstract

Abstract. The application of spatial modelling to epidemiology has increased significantly over the past decade, delivering enhanced understanding of the environmental and climatic factors affecting disease distributions and providing spatially continuous representations of disease risk (predictive maps). These outputs provide significant information for disease control programmes, allowing spatial targeting and tailored interventions. However, several factors (e.g. sampling protocols or temporal disease spread) can influence predictive mapping outputs. This paper proposes a conceptual framework which defines several scenarios and their potential impact on resulting predictive outputs, using simulated data to provide an exemplar. It is vital that researchers recognise these scenarios and their influence on predictive models and their outputs, as a failure to do so may lead to inaccurate interpretation of predictive maps. As long as these considerations are kept in mind, predictive mapping will continue to contribute significantly to epidemiological research and disease control planning.

Influenza A H5N1 and H7N9 in China: A spatial risk analysis

“…SDMs are most commonly used in the fields of biogeography, conservation biology and ecology to identify environmental conditions which relate to species occurrence, estimate current species distribution and predict species distributions in new areas or under new environmental conditions [10,11]. In recent years, SDMs have been increasingly used in the public health context [12,13], particularly with respect to infectious diseases that involve vector species such as mosquito-borne diseases [14,15] and tick-borne diseases [16]; these diseases are markedly influenced by subtle changes in climatic conditions which determine vector distribution. In this study we created SDMs using Maxent for both H5N1 and H7N9.A small set of previous studies have used SDMs to identify the set of high risk areas to target for control for both H5N1 [17-23] and H7N9 [24][25][26][27][28][29][30].…”

Data-driven and interpretable machine-learning modeling to explore the fine-scale environmental determinants of malaria vectors biting rates in rural Burkina Faso

“…However, careful attention should be given because of well-known problems linked to predicting beyond the model sampling locations or range of values (e.g. overfitting) [ 107 , 108 ]. The scalability of our models to places with similar landscape and weather dynamics as the Diebougou area could be tested by collecting similar entomological data in another health district and comparing these ground-truth data with predictions generated by the models.…”

To Reduce the Global Burden of Human Schistosomiasis, Use ‘Old Fashioned’ Snail Control

“…Some indirect sampling methods might become cost-effective with additional refinement. For example, species distribution modeling [47, 48] encompasses various methods to correlate species occurrences to underlying habitat variables, such as temperature, rainfall, vegetation cover, etc. This technique could help generate maps that predict schistosomiasis transmission hotspots using readily available data, like land features and environmental variables [49].…”