Back in 2016, four years before a pandemic saw the world grind to a halt, the United Nations Environment Programme (UNEP) was sounding the alarm on zoonotic diseases, identifying them as a key emerging issue of global concern.
Now, according to the World Health Organization, around one billion cases and millions of deaths each year are the result of zoonoses, in which pathogens jump from vertebrate animals to humans. And of the 30 novel human viruses that have been identified in the last three decades, a massive 75% originated in other animals.
But scientists at the University of Montreal believe their new artificial intelligence modeling has the capacity to highlight and predict emerging viral "hotspots" to watch, which could get the jump on likely breakout animal-to-human infections and, ideally, prevent anything like COVID-19 from happening again.
The algorithm, which took researchers three years and 10,000 hours of computing, was able to identify 80,000 new potential interactions between viruses and hosts, and where in the world they’re of most concern.
“We had been working on this project from the first few months of 2020, before the pandemic took off,” said Timothée Poisot, a professor in the Department of Biological Sciences at the University of Montreal.
Through machine learning, rather than manually making links in data, the algorithm was able to assess thousands of mammal species and thousands of viruses and work out all the viable combinations.
“The basic problem is that we are only aware of between one and two per cent of the interactions between viruses and mammals,” Poisot said. “The networks are scattered and there are few interactions, which are concentrated in just a few species. We want to know which species of virus is likely to infect which species of mammal, so we can establish which interactions are most likely to occur.”
The team used the largest open dataset, CLOVER, which described 5,494 interactions between 829 viruses and 1,081 mammalian hosts, a majority of which focused on wild animals, as well as several other datasets, including the Host-Pathogen Phylogeny Project (HP3), Enhanced Infectious Diseases Database (EID2) and the Global Mammal Parasite Database V2.0 (GHMPD2).
“Some of the data sets we had were older: they contained out-of-date names for particular species, or they had errors because the data had been entered by hand,” Poisot said of the time-consuming process that was required for the machine learning. “After that, the main task was to determine the level of confidence we had in the model’s ability to make predictions."
The researchers then focused on 20 viruses that were deemed ones of concern and that had the potential to spill over to humans.
“We had a lot of discussions on the team, because at first some of the results seemed strange to us,” said Poisot, who was surprised to see the mice-linked Ectromelia virus identified as one to watch. “We were skeptical, but when we searched the literature, we found there had been cases in humans.”
The researchers were also able to pinpoint regions through the model, something that could help scientists pursue viral and vaccine research in a more targeted way.
“Our model makes spatial predictions, but more precisely, the model indicates specifically in which group of mammals and in which location certain types of virus are likely to be found,” said Poisot.
The results showed two areas of specific interest: the Amazon basin, where virus and host interaction are more original and new interactions are most likely to be seen; and Sub-Saharan Africa, where the algorithm identified new hosts likely to carry zoonotic viruses.
“We are really shifting the places where we need to go and study mammals to discover new viruses,” Poisot explained.
While zoonotic pathogens can take many forms – bacterial, parasitic, viral – their prevalence is expected to be increasingly more common as human and non-human animals continue to occupy more of the same space.
The team hopes its model can not only inform new starting points for research but offer real-world surveillance. The next step would be to take this AI to the next level and include more microbiological, immunological and ecological mechanisms, for a more complete look at a global virome.
“The algorithm takes the network we already know, and projects it into a new space, a bit like shadow theater: it casts light on interactions in a new way," said Poisot. “We now know which species to monitor, where and for what type of virus.”
The research was published in the journal Patterns.
Source: University of Montreal