We’re often asking ourselves, “What’s the weather going to be like tomorrow?” But what if we could ask, “What’s my health going to be like in 2035?” Well, dear reader, it seems Artificial Intelligence might just have the answer. Scientists are now telling us that AI can predict future health problems over a decade in advance, much like a meteorologist forecasting a 70% chance of rain. Only, instead of rain, we’re talking about the likelihood of over 1,000 different diseases. Fascinating, isn’t it?
The technology, dubbed Delphi-2M, isn’t some crystal ball giving exact dates for a heart attack. Instead, it estimates the probability of developing 1,231 diseases. Professor Ewan Birney, the interim executive director of the European Molecular Biology Laboratory, is quite chuffed about it, stating, “We can do that not just for one disease, but all diseases at the same time – we’ve never been able to do that before. I’m excited.” And frankly, so should we be.
This clever bit of kit uses similar tech to those chatty AI chatbots like ChatGPT. But instead of predicting words, Delphi-2M has been trained to find patterns in anonymous medical records so it can predict what comes next and when. It initially cut its teeth on anonymous UK data from over 400,000 people via the UK Biobank research project, which includes everything from hospital admissions and GP records to lifestyle habits like smoking. They then tested its predictions against other Biobank participants and, rather impressively, 1.9 million medical records in Denmark. And guess what? “It’s good, it’s really good in Denmark,” says Prof Birney. “If our model says it’s a one in 10 risk for the next year, it really does seem like it turns out to be one in 10. Now, that’s what we call precision.
Naturally, the model is a bit of a whizz at predicting conditions with clear progression, like type 2 diabetes, heart attacks, and sepsis, rather than those pesky, more random infections. The vision, you see, is to use this AI model to identify high risk patients before a disease fully takes hold. Imagine interventions like targeted medicines or personalised lifestyle advice, for example, advising someone at high risk of liver disorders to cut back on the vino more than the average Joe. It could also revolutionise disease-screening programmes and help hospitals anticipate demand years in advance, like how many heart attacks Norwich might see in 2030, allowing for better resource planning.
Professor Moritz Gerstung, head of the division of AI in oncology at DKFZ, the German Cancer Research Centre, rightly calls this the beginning of a new way to understand human health and disease progression. He believes these generative models such as ours could one day help personalise care and anticipate healthcare needs at scale.
Now, before we get too carried away, it’s important to stress this is still very much research. It needs refining and testing before it is used clinically, as detailed in the scientific journal Nature. There are also potential biases, as the UK Biobank data largely represents people aged 40 to 70. But fear not, the model is being upgraded to incorporate more medical data, including imaging, genetics, and blood analysis. Prof Birney anticipates a decade long journey, similar to that of genomics in healthcare, before this kind of predictive modelling becomes routine. This collaborative effort from the European Molecular Biology Laboratory, DKFZ, and the University of Copenhagen is, as Professor Gustavo Sudre from King’s College London notes, a significant step towards scalable, interpretable, and most importantly ethically responsible form of predictive modelling in medicine.
But wait, there’s more! While scientists are cautiously optimistic, Bill Gates has thrown a rather large spanner in the works with his prediction that AI will replace doctors and teachers by 2035. According to reports, the Microsoft co founder envisions a “Free Intelligence Era” where great medical advice and great tutoring become free and commonplace as AI handles specialised human skills. He says, “intelligence is rare, you know, a great doctor, a great teacher. And with AI, over the next decade, that will become free, commonplace.” This, he suggests, is a transformation akin to computing becoming virtually free, but now applied to human expertise.
These capabilities promise faster diagnoses, early disease detection, and more personalised treatment plans. However, most healthcare professionals and researchers wisely emphasise that AI will augment rather than replace doctors, serving as a powerful collaborative tool that enhances human medical expertise. So, while we might not be seeing robot doctors operating on us just yet, the future of healthcare is certainly looking a lot more intelligent.
