Hi everyone,
which analysis can be used for disease distribution modeling (DDS) with polygonal data (each polygon has number of disease incidence as a field and it may be as vast as 40,000 km2 and our work is in country scale level (more than 1.5 m km2)).
Is it possible to convert each polygon to pint (like centroid) and then use it in ecological niche models like MaxEnt or other types of models)?
There are number of methods to analyze areal disease data. You can find number of research papers available online. The choice of methods, in my view, depends not only to the type of disease data; whether aggregated in polygon or available with point but also with research problem. I have used semi parametric geographically regression model to assess the spatial variation and associated risk factors of (https://www.ncbi.nlm.nih.gov/pubmed/30182200) using the areal aggregated dengue data. I suggest GeoDA website (https://spatial.uchicago.edu/software) to have a look on some available spatial analysis methods (such as spatial regression Lag, spatial error model, geographically weighted regression etc.) for the spatially aggregated data.
Regarding the queries about whether it is possible to convert each polygon to pint (like centroid) and then use it in ecological niche models like MaxEnt or other types of models)? I don’t suggest it do so. However, you may also find that type of literature analyzed using the maxent based on the centroid of polygon of disease reported areal unit (https://www.sciencedirect.com/science/article/pii/S0143622811001275). Therefore all depends on research problems and objectives of the research.
Hello Mojtaba,
If you are able to get the point data, I suggest you use species distribution modelling or ecological niche modelling to estimate the distribution range. If you got the occurrence data and environment data, there are many method to perform the analysis. You can use sdm, biomod or dismo package in R platform. As a simple alternative, you can use Maxent along to do that.
Cheers,
Jie
Dear Mojtaba,
just a short supplement to the answers of Bipin and Jie: you can use ENM not only in the centroids but also using the original polygons (if your auxiliary/environmental data characterize the entire polygons, such as population, mean GDP, mean temperature etc.). Or if you have raster/gridded auxuliary dataset, you can 'downscale' your polygonal data to the raster cells. If you use a simple ovelapping (extract values to raster cells), then this won't increase the resolution of your data, but helps you to use the dependent and independent variables in one model.
HTH,
Ákos
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping