By Katie Joy Sanchez
- Researchers are amassing databases which contain data on animals and viruses.
- They are attempting to use machine learning to predict what potential spillovers are the most likely to happen
- They are also using machine learning algorithms to predict where these spillover events may happen.
Although for years, biologists have been concerned about virus spread from animals to humans, they have just recently begun to apply machine learning as a methodology of early virus spillover identification.
For example, some machine learning programs are attempting to learn what ecological and biological factors of the animal hosts can lead to a higher likelihood of viral spillover. Such factors include information about various bats, their habitat, their wingspan, and their diet. This information is then inputted, and a predictive model attempts to determine the likelihood that certain animals associated with those factors might harbor a specific virus.
Researcher have also tried another approach: they have analyzed the viruses themselves and their specific features such as their genomes. Then, the researchers have applied machine learning to this information to help to identify and potentially predict the features of a certain virus that may be more likely to affect humans.
Some machine learning algorithms have actually produced a list of viruses that may likely affect humans in the future. One of the most concerning at the moment is mousepox, which already has had a few cases of spillover, but not enough to be considered a threat. Still, there was a case of an outbreak of mousepox in humans in China that has largely been forgotten which occurred in 1987. Throat swabs taken during the outbreak were used and have been compared with machine learning and ranks very highly as a virus of concern.1 Historically, there have been about 250 human disease viral spillover events, the most impactful of which in recent memory being the coronavirus.2
Climate Change Affecting Spillover Location
Since the coronavirus originated in bats, this is an increasing concern, as the public is now well aware of the effects which viruses can have on global society. However, there perhaps is less knowledge about the impacts which climate change is having on the increase of viral spillover.
According to a new study, thousands of viruses may begin to jump into humans due to the effects of climate change. Scientists at Georgetown who are conducting the machine learning project are also building a computer model of these new viral spillovers. They are projecting how a warmer world may impact animal movement from one region to another. The changes in ranges of habitat for these animals may increase their contact with humans and lead to more viruses jumping the species barrier. By using inputs on 3,139 species in a machine learning prediction model, the computers will be able to track where it is more likely that a spillover may occur.3
Scientists have unveiled a database that has approximately half a million pieces of data. This database is called VIRION. It will allow for the development of further research into viruses that may lead to new pandemics. It will also aid in the creation of more machine learning algorithms and prediction models to track these viral risks.
- Genome Sequence of Erythromelalgia-Related Poxvirus Identifies it as an Ectromelia Virus Strain Mendez-Rios JD, Martens CA, Bruno DP, Porcella SF, Zheng ZM, et al. (2012) Genome Sequence of Erythromelalgia-Related Poxvirus Identifies it as an Ectromelia Virus Strain. PLOS ONE 7(4): e34604. https://doi.org/10.1371/journal.pone.0034604
- Zimmer, Carl. “Which Animal Viruses Could Infect People? Computers Are Racing to Find out.” The New York Times. The New York Times, April 27, 2022. https://www.nytimes.com/2022/04/27/science/pandemic-viruses-machine-learning.html.
- Zimmer, Carl. “Climate Change Will Accelerate Viral Spillovers, Study Finds.” The New York Times. The New York Times, April 28, 2022. https://www.nytimes.com/2022/04/28/science/climate-change-virus-spillover.html.