Fighting Disease X: AI & the next Pandemic

Listening to the big health organisations, the question doesn’t seem to be if there will be another pandemic; it’s when. ?

It’s a scary thought. No one wants another lockdown, no one wants that horrible announcement of weekly death tolls or an overwhelmed healthcare system. But; would that necessarily be the case next time around? A lot’s changed since 2019. ?

Experts have been warning for years about “Disease X,” the hypothetical (but supposedly inevitable) pathogen that will trigger the next global health crisis. Now, whilst I'm absolutely no virologist, I’ve recently been on a bit of a health-tech kick, and after reading a great (but all too brief) BBC article about AI’s role in the next pandemic, ended up going down my all-too-familiar late-night rabbit hole frenziedly trying to learn as much as possible about AI and pandemics. ? ? Unfortunately, a lot of it was ring-fenced behind paid-for Scientific journals and what was easily available was absolutely not digestible for an audience (myself included) who aren’t deeply literate in p-values and the finer points of protein folding, so – I made it my mission to wade in and try and put together a piece that made some sense of all of the great work that’s being done to utilise the massive power of AI in fighting the next pandemic, and make that as accessible and understandable as possible. ?

So, what did I find? Well, I’ll front-load it – it looks exciting. Like, ridiculously so. I genuinely think in the future we’re going to be able to attribute the survival and recovery of hundreds of thousands of people, if not millions – to the advances made possible by AI. That said, there’s a lot of nuance in the process that’ll lead us there, and it’s not monolithic, there’s thousands of ways in which AI is being utilised to potentially save lives, some of them are going to be winners, others won’t be.

"It’s not monolithic, there’s thousands of ways in which AI is being utilised to potentially save lives, some of them are going to be winners, others won’t be."?

They say an ounce of prevention is worth a pound of cure. If that’s the case, we’re off to a flying start. One of AI’s standout abilities is predicting outbreaks. Platforms like EPIWATCH trawl through mountains of data, everything from social media posts to public health bulletins, looking for tell-tale patterns that might hint at a brewing epidemic. By catching these signals early, governments could act before things spiral out of control. Imagine being able to ramp up testing or close borders at just the right moment, the lives that could save.

Of course, it’s not all smooth sailing. The system is only as good as the data it’s fed, and not all countries have the resources—or the political will—to provide accurate, timely information. Gaps in data can quickly become gaps in AI’s predictions, and in a health crisis, those gaps can cost lives. Still though, with all the usual caveats, this looks incredibly promising. ?

Okay, so AI can help us react and detect outbreaks early, but how about dealing with an already established outbreak? Well, AI is also shaking up how we discover new drugs, which is more important than ever as antimicrobial resistance continues to rise. Traditional drug discovery can take years, but machine learning?can scan through billions of potential compounds in a fraction of the time, identifying promising candidates much faster. Tools like these have already led to breakthroughs, such as halicin, an antibiotic that has proven effective against some drug-resistant bacteria. Researchers are also using AI to design entirely new types of drugs, predicting how they’ll interact with pathogens and even identifying their mechanisms of action. This is no small feat - it’s?catapulted the rate of discovery forward, with previous, ‘brute force’ methods being hundreds of times slower by comparison. This one’s an absolutely huge win, and one with an already-proven record, so can’t simply be attributed to the usual cycle of tech-industry hype. Definitely the kind of thing that gets me really excited about all the breakthroughs we’re likely to see over the next couple of decades. ?

Speaking of both speed and tried-and-tested results, AI has also already proven its worth in vaccine development. During the COVID-19 pandemic, AI tools helped compress what would typically take months into just days. They also allowed researchers to simulate how vaccines might perform against potential new variants, giving them a head start in adapting their designs. Stuff like the Omicron variant was predicted ahead of time, which just sounds crazy to me! This isn’t just useful for global emergencies. AI could be a game-changer for tackling diseases that are overlooked in the media relative to how deadly they are, like malaria or tuberculosis, which still claim millions of lives each year in the global south. ?

"AI tools helped compress what would typically take months into just days."

Diagnostics is another area where AI is stepping up. Traditional methods can take days, even weeks, to confirm results. AI, combined with cutting-edge tools like CRISPR, is helping to create faster, cheaper, and more portable diagnostic kits for diseases like malaria and Zika. Some AI models can even predict antimicrobial resistance in just 24 hours - a huge improvement over older methods. But, as impressive as this sounds, it’s not all perfect. Poor-quality or biased data can lead to errors, and the so-called "black box" problem - where AI makes predictions without explaining how - can undermine trust in the technology, especially with trust in medical establishments already in short supply after the spread of conspiracy theories and disinformation during the COVID pandemic. ?

Speaking of trust, the ethical side of AI is a major talking point. Patient data is incredibly sensitive, and breaches can be catastrophic. There have been high-profile cases, like Google DeepMind’s alleged use of NHS patient data without consent, that highlight just how serious these issues can be. To tackle this, researchers are exploring techniques like federated learning, which allows AI to improve without compromising individual privacy. However, ensuring public confidence will also require transparency in how AI systems are designed and deployed. Trust is hard-earned, but in healthcare, it’s non-negotiable.?

Another challenge AI faces is bias. AI models can only be as unbiased as the data they’re trained on, and if that data underrepresents certain groups, the results can be unfair. For example, models trained on predominantly white populations may not work as well for other ethnicities, leading to unequal outcomes. While solutions like multi-ethnic datasets and explainable AI are helping to address this, diversity in development teams and training data is essential to ensure AI benefits everyone equally.?

Of course, it’s not just technical challenges that need solving. Integrating AI into the day-to-day running of hospitals and clinics is no small task. Many AI tools are still experimental and haven’t been tested enough in real-world settings. For AI to truly make a difference, it needs to blend seamlessly into existing workflows, rather than adding more stress to already overworked healthcare professionals. Education and training for medical staff are also vital, so they’re equipped to use AI effectively. Some medical schools are starting to introduce AI basics into their curriculums, but there’s still a long way to go.?

"Education and training for medical staff are also vital, so they’re equipped to use AI effectively."

AI isn’t limited to pandemics or infectious diseases either. It’s helping to develop diagnostic tools for cancer, simulate drug delivery systems, and even create less invasive medical procedures, like video capsule endoscopy. These innovations could lead to faster diagnostics, fewer complications, and lower costs, making high-quality healthcare more accessible to all. ?