Most of the time we all walk around in a little bubble, a defence system that can spot a threat and neutralise it before it has a chance to harm us. That’s the wonder of the human immune system, and it’s only when we get ill do we become aware that it’s there at all.
What if we could give the planet an immune system just like ours? A silent network of satellites and supercomputers quietly keeping track of anything that could cause the next pandemic; primed to sequence the culprit and capable of rolling out vaccines and treatments the second someone presses the right button. That’s the vision of global technology expert Dr David Bray, who believes that we need to build “an immune system for the planet”.
“I’m talking about a dynamic system that learns to respond to what’s present in our world,” Bray says. “If you think of our world as an organism and we are parts of that organism, then what do we need to do, much like the immune system of our bodies, to detect that there’s something going on that’s not healthy?”
Bray is a Distinguished Fellow at the Stimson Center, a not-for-profit based in Washington, DC, that carries out research to solve big, real-world problems using technology. He talks at the speed of his brain – fast – probably because he has a lot to cover. Starting at age 15 working for the Department of Energy, he’s used satellites to spot forest fires, built computer models of HIV/AIDS, got a PhD in “organisational responses to disruption”, briefed the CIA on bioterrorism and carried out independent analysis on the Afghanistan situation for the Obama administration.
It’s not just big problems that Bray is interested in solving; it’s the near-impossible ones. “You tell me it’s never been done before, you tell me it’s impossible and I’m like, ‘okay, that’s what I want!’” he laughs. Like preventing pandemics? Yes, he says, though he explains that the goal is not necessarily preventing them, but detecting and reacting to them as early as we can. What happened with COVID, he says, is that we burned up too much time before it was taken seriously.
But can we really prevent the next pandemic by throwing technology at the problem? University of Glasgow epidemiologist Prof Sarah Cleaveland is sceptical. “I just think we have to temper our expectations about some magic technological solution,” she says. Though we are getting closer to being able to look at a virus, or its genome, and predict whether it’s going to be a problem for humans, that’s “not possible” right now, Cleaveland adds. Not possible: all the encouragement Bray needs.
Read more about pandemics:
How AI could detect the next outbreak
The planetary immune system is an idea Bray proposed to the US Defense Advanced Research Projects Agency (DARPA) in 2013. DARPA was busy with Iraq and Afghanistan, so he was told to come back in “about seven or eight years, which, of course…” Bray nods to the COVID situation again. “Well, here we are now.”
In 2021, he outlined a new scheme for the concept, which is perhaps easiest to understand in the context of how our own immune systems work. Human immune systems detect disease carriers through a diverse array of receptors on living cells, spread throughout every organ and tissue. Likewise, Bray imagines a system that picks up on potential threats via a global detection network, one that would include sensors spread through the air, soil and water, as well as artificial intelligence (AI) programs trained to spot weird blips in patterns of human behaviour and economic markets that might point towards infectious interference.
With a potential threat identified, the system’s vast processing power would be thrown at sequencing it, characterising its interactions with human cells and using that information to figure out how to stop it. In an ideal world, this would all happen in machines, with as little input from us as possible, reducing the response time from years to three or four weeks.
Perhaps it sounds a little out there. But taking it piece by piece, we’re already doing a lot of it, and it just needs tying together. Take weird blips, for instance – something Bray specialises in. Between 2000 and 2005, Bray was the IT chief for the Bioterrorism Preparedness and Response Program at the US Centers for Disease Control and Prevention, where part of his job was looking for quirks in data that might act as early indicators of a bioterrorism attack.
If the President was speaking at the Super Bowl, his team would start three days ahead, establishing baseline levels of over-the-counter drug sales, emergency calls and school absences. “If all of sudden I [saw] a spike when the President [was] speaking, that could be a bioterrorism event,” he says.
Another example is garlic in China, where it’s considered a remedy for all ills. The price shot up when SARS struck the Guangdong province in 2002, and again in early 2020, as COVID was taking off. In 2002, the garlic price hike, combined with the increasing number of cars showing up in hospital car parks, gave Bray and his colleagues a hint that something was going on. “We knew about it five and a half months before it hit the world stage,” he says. Artificial intelligence could detect these sorts of oddities, according to Bray’s concept, forming a key component of an early alert system.
This surveillance for strangeness is just one small part of the scheme. The planetary immune system would double-down on disease by combining different detection methods.
Tracking disease in sewage
Next up: the sewers, one of the places Bray suggests we should be looking for pathogens. Here again, there’s a precedent. In 2013, sewer sampling caught a silent epidemic of a poliovirus in Israel, triggering a vaccination campaign that helped to avert a more serious outbreak of polio. Now, across Europe and the US, wastewater testing for COVID-19 is helping scientists to get an unbiased picture of how the virus spreads.
Italian researchers, for example, retrospectively analysed samples of raw sewage collected before the first wave officially began, detecting traces of the virus in samples from three Italian cities in December and January, weeks before the first non-imported case was reported in late February. And according to Prof Dragan Savic, CEO at KWR Water Research Institute in Nieuwegein, the Netherlands, we’re already looking at levels of new variants.
Sewage, Savic explains, provides a window on literally everyone’s business. So you don’t have to wait for mass testing to get up and running, or worry about whether people are testing when they should be. “There are people who don’t have symptoms, there are people who are recovering – who will not go for testing – and there are limited tests,” he says. “But we all go to the toilet, so when you think about it, that’s unbiased sampling.”
So our wastewater can give us a heads-up when an emerging disease first infects a new area. With AI, the same information could also be combined with other data sources, like weather patterns or gatherings of people, to help reveal how an infectious agent is spreading and plan allocation of resources.
Another key component of Bray’s system is a biosensing network that he envisions being embedded into plants and animals. In his 2021 scheme, he describes transmissions from these biosensors feeding into a supercomputing network. Such biosensors aren’t fantastical; they are promising and, in some cases, already commercially developed devices (think glucose sensors for diabetes) based on biological components. They’re set to make a big impact as “low-cost sensing systems that can be deployed in the field”, according to synthetic biologist Prof Paul Freemont, who is developing biosensors at Imperial College London.
Freemont’s biosensors are fluorescent beads that lose their fluorescence when they contact proteases, which are enzymes used by a wide variety of organisms. Many viruses, from HIV to herpes, use proteases to replicate. Freemont has tailored them to detect different targets, including proteases from a plant virus, as well as from Schistosoma mansoni, a parasitic worm. The technology uses components that the researchers suggest could be modified to detect different pathogens, perhaps in portable devices in regions where the risk of animal diseases crossing over into humans is high.
However, embedding biosensors into plants and animals may be taking the idea too far for Freemont. He isn’t convinced that biosensors would help us get ahead of a new disease. “My question is: how can we embed sensors to detect things that we know nothing about?” he says. “It could be useful for existing targets, for sure, and for emerging targets that we know of, but the problem is something completely new. Would you have enough time to redesign that system and get it back into the wild?”
It’s a good question. Our immune systems don’t just respond to bacteria and viruses that they’ve met before. They detect and protect us against any foreign invader, using receptors that distinguish between ‘self’ (us) and ‘non-self’ (invaders). It’s hard to envisage how we might replicate this system on a planetary scale. But we need to be able to work out what, from the multitude of unknowns, might pose the biggest threats. This, Bray says, will be “the most novel part” of the system.
Dr Kevin Esvelt, a biochemist at the Massachusetts Institute of Technology, has his own ideas about how we should detect new and dangerous pathogens. His theories are aimed primarily at dealing with what he calls “the deliberate variety” of pandemic – to his mind a more difficult problem than a naturally occurring one. Why? Because, he says, bioterrorists have to be expected to design weapons to evade any defences that we might set up. We therefore need a more open approach in order to detect this stuff; one that would catch not just new variants of old diseases, but something the like of which we’ve never seen.
In a pre-print paper published on arXiv in August 2021, Esvelt proposes a global Nucleic Acid Observatory that would sequence the genomes of everything we encounter in the sewer space and beyond. But how would this approach help us to pinpoint those organisms we need to be particularly wary of? As the paper outlines, the real hallmark of any organism with pandemic potential is unbridled growth, which would be seen as an explosion in levels of its genome sequences.
“Any serious threat must by definition grow exponentially,” Esvelt explains. “This should allow us to detect anything potentially problematic, be it an invasive pest, a crop blight, a novel pandemic virus unrelated to anything we’ve seen before, or even something totally new that [an] adversary has devised.”
Perhaps something like this observatory could form a key part of the immune system for the planet? Freemont, for his part, says it’s a “compelling” proposition for solving the problem of detecting entirely new pathogens. Yet to do all of this, we’d need armies of sample collectors to mail samples back to the lab every few days. Compared to the seamless detection system that Bray imagines, this does seem rather laborious.
But Esvelt argues it would be worth it. “Doing this for next-day sequencing is a tiny cost relative to that of the sequencing itself, and losing a day in exchange for reliability is an excellent trade-off,” he says.
Sequencing won’t tell us everything though. Any candidates with pandemic potential would have to be probed in exquisite, molecular-level detail, not just to work out whether they’re real threats, but also how we might tackle them. “That’s going to involve some research,” Bray admits.
At Washington State University, a project called Deep VZN is already aiming to do something similar. Deep VZN will characterise thousands of viruses from wild animals across 12 countries, focusing on the families of viruses that spawned COVID-19, Ebola and measles.
Project leader Dr Felix Lankester says his team plans to probe the molecular structure of those whose genomes are “unknown”, looking for a lock-and-key type fit between molecules on the surface of viruses and receptors on human cells. They will also use an algorithm to assess the likelihood of the viruses spilling over into humans and look at whether these viruses could evade or inhibit our immune defences. “This information will be made available for the design and development of countermeasures, such as new vaccines and diagnostics,” says Lankester.
In the future, Bray anticipates a lot more of the work being done by machines. With AI programs like DeepMind’s AlphaFold already making huge strides in predicting the 3D structures of proteins, such thinking might not be such a stretch. So, if you leave aside the difficulty of predicting how symptoms might play out in real humans (before any have actually been infected), all of this might sound quite feasible, eventually.
It’s when we get to Bray’s ideas about how we should respond to such threats that things get a bit wild. He would have us rolling out machine-designed vaccines at the touch of a button and even suggests distributing antibodies in drinking water, a step that would seem less drastic if we were talking about something more dangerous than COVID. Which he is.
“We’re not talking about a slow-spreading pandemic like we have now. It’s one where you would have an hour or less to survive,” he says, suggesting that the first place to trial this would be in countering the use of bioweapons on military battlefields.
The question is: will policymakers be persuaded to think ahead on pandemics? Even post-COVID, warnings of epidemiologists are still ignored, Cleaveland points out. Hendra virus, a bat disease that can be fatal to horses and humans if it spills over, is a case in point.
In 2020, Australian and US researchers warned of a high risk of spillover during the coming winter, but to no avail. “For me that was a really fascinating insight,” she says. “We have a vaccine. This is a fatal disease. We had a good indication that there was a risk coming and still, people didn’t respond.” Luckily, things turned out okay, but only because the bats were kept at bay by an unexpected crop of fruit.
So if policymakers aren’t willing to act, who is going to invest in the immune system for the planet? “Can we find a coalition of the willing?” asks Bray. It’s always going to be a hard sell because if it works, nothing will happen – preventing a pandemic just looks like a colossal waste of cash. Unless, maybe, you’re still reeling from the financial fallout from the last one.
Read more about COVID-19: