Machine learning can pinpoint rodent species that harbor diseases andgeographic hotspots vulnerable to new parasites and pathogens. So reports a new study in the Proceedings of the National Academy of Sciences led by Barbara A. Han, a disease ecologist at the Cary Institute of EcosystemStudies.
Most emerging infectious diseases are transmitted from animals to humans, with more than a billion people suffering annually. Safeguarding public health requires effective surveillance tools.
With University of Georgia Odum School of Ecology colleagues John Paul Schmidt, Sarah E. Bowden, and John M. Drake, Han employed machine learning, a form of artificial intelligence, to reveal patterns in an extensive set of data on more than 2,000 rodent species, with variables describing species' life history, ecology, behavior, physiology, and geographic distribution.
The team developed a model that was able to predict known rodent reservoir species with 90% accuracy, and identified particular traits that distinguish reservoirs from non-reservoirs. They revealed over 150 new potential rodent reservoir species and more than 50 new hyper-reservoirs - animals that may carry multiple pathogens infectious to humans.
http://phys.org/news/2015-05-future-infectious-disease-outbreaks.html
[Abstract]: http://www.pnas.org/content/early/2015/05/14/1501598112
[Source]: http://www.caryinstitute.org/newsroom/forecasting-future-infectious-disease-outbreaks
(Score: 1, Insightful) by Anonymous Coward on Wednesday May 20 2015, @09:13AM
I don't think it is really a prediction if the species were already known.
(Score: 2) by sudo rm -rf on Wednesday May 20 2015, @12:45PM
Maybe the wording is a bit weird, but a model *does* predict an outcome, and when the result includes already known species to a high accuracy, it can be assumed it is well enough to extrapolate this accuracy to the other 200 reservoir species. The model allows you to narrow the search.
(Score: 0) by Anonymous Coward on Wednesday May 20 2015, @01:56PM
Perhaps, it says they used 80% of the data for training and 20% for testing though. And anyone who has done this type of thing knows you try out different approaches to see how good of results you can get. I think you have to wait for new data before getting excited. Kudos to them for putting the prediction out there though.