It's been nearly two decades since a coronavirus capable of wreaking havoc on our bodies threatened global pandemic. Last year, that threat was made good with the emergence of a deadly new version, dubbed SARS-CoV-2.

With the world still reeling from the devastation of COVID-19, the question of if – and when – we'll see yet another member of this family pop up in the near future has the attention of researchers. Their findings should have us on high alert.

A recent investigation has employed machine learning to predict which mammals could host multiple strains of coronavirus, allowing the pathogens to mix and piece together the makings of a next-gen COVID.

The numbers hint at close to a dozen times more coronavirus-host associations than estimates based on observations alone.

Alarmingly, they also found more than 30 times the potential hosts that could harbour SARS-CoV-2 and allow it to recombine into something uglier; and over 40 times the number of species previously suspected to host a handful of coronavirus subgenera.

The study, carried out by researchers from the University of Liverpool in the UK, can't tell us where or even when future pandemics will arise, but it does suggest our current models on the coronavirus ecosystem are in serious need of an overhaul.

While the name of this family of spiky viruses is now synonymous with 2020's shut-downs and high death toll, most of its members are relatively benign. It's even likely you've experienced infections as a seasonal sniffle in the past.

There are four genuses making up the coronavirus family, simply titled alpha, beta, gamma, and delta, which variously infect a wide variety of mammals and birds.

Of all of these, just alpha and beta coronaviruses contain variants that infect humans, with the latter of those harbouring severe acute respiratory syndrome (SARS) strains.

While we love sorting biology into clear categories, nature isn't quite so discriminating.

Viruses regularly shuffle genes and recode their genome as they infect hosts, chancing on clever new methods for unlocking cells, evading eviction from the immune system, or even jumping to new animals, giving rise to what we call new species and strains.

Just to make it all more confusing, members of different groups can congregate inside the same tissues of a host and swap those handy new tools, giving rise to ever more powerful combinations that take the virus's spread to a whole new level.

Keeping track of this swap-meet is no easy feat. Researchers are still trying to work out the exact origins of SARS-CoV-2, with evidence currently pinning it on bats.

Thanks in part to the amazingly robust immune systems, many bat species can harbour a number of viruses comfortably for long periods, giving them all a chance to mix-and-match their genes.

From there, a leap into a human host simply requires a chance encounter, either directly or through an intermediate host, such as a pangolin brought into civilisation for its meat.

Whether this was simply bad luck or an inevitability depends entirely on the frequency of recombination events. And the answer to that question depends on knowing a thing or two about the diversity of the viruses, their potential hosts, and the circumstances under which they meet.

Virologists aren't completely in the dark on these facts, but are also aware that they've barely seen the tip of an epidemiological iceberg when it comes to viruses quietly jumping from species to species, especially in the wild.

In this case, computerised algorithms were designed to find patterns among three different but complementary perspectives – genomic features among the virus's family tree, traits of hundreds of potential mammalian hosts, and characteristics of the virus-host network.

The results reveal the breadth of the landscape open to members of the coronavirus family for sharing their secrets. And with SARS-CoV-2 still circulating, it's clear that there is huge potential for combinations to form that could quickly get out of hand.

That landscape is looking pretty vast. Observations tell us there are only four non-human mammals known to be able to host both SARS-CoV-2 and one other coronavirus.

When the researchers' data is taken into account, there's a whopping 126 SARS-CoV-2 hosts and 2,544 total unique interactions that could teach SARS-CoV-2 how to return with a vengeance.

"Any of these SARS-CoV-2 hosts that are also hosts of other coronaviruses are potential recombination hosts in which novel coronaviruses derived from SARS-CoV-2 could be generated in the future," the researchers explain.

It isn't enough to tell us the exact odds of a new COVID emerging in the near future. But whatever those odds might be, the possibility of another SARS virus outbreak occurring in the future isn't something we ought to be gambling on.

This research was published in Nature Communications.