Although the devastating Ebola outbreak in West Africa has subsided and is no longer demanding the public’s attention, researchers continue to examine the situation in hopes of preparing for–and preventing–future epidemics. Now, a group has applied artificial intelligence and machine learning to analyze data and learn about the role of bats in spreading the virus.
To start, experts from the University of Georgia, Massey University and the University of California built a profile of filovirus-positive bats from 21 species known to harbor viruses such as Ebola. Then, using 57 variables including diet, migratory pattern and species density, an algorithm was pored over data to identify virus-spreading species with 87% accuracy, according to a release.
The technology represents a shift, Massey University’s David Hayman explained, because it allows scientists “to move beyond our own biases and find patterns in the data that only a machine can.”
Further, researchers can “forecast risk” based on data analysis instead of looking at past outbreaks to make predictions, Hayman explained.
Through the model, the team found that traits such as early maturity and large offspring are associated with filovirus-positive bat species. The group also found that they are found in more places than previously known, ranging from sub-Saharan Africa to Southeast Asia and Central and South America.
For future work, University of Georgia’s John Drake said an “outstanding question” would be to examine, given the new information, why there are so many more virus-spillover events in Africa compared to other geographies.
– here’s the release