Imagine waking up one morning, checking your phone, and instead of only seeing whether it might rain, you also see a forecast of your future health. “70% chance of developing type 2 diabetes in the next 10 years,” or “10% chance of a heart attack within five years.” Sounds like science fiction? Not anymore.
Scientists have now developed an artificial intelligence (AI) model that can predict people’s health risks years—sometimes decades—into the future, just like a weather forecast.
The AI model, called Delphi-2M, has been trained on medical records from hundreds of thousands of people. Much like AI chatbots (such as ChatGPT) learn patterns in language, Delphi-2M learns patterns in health data—hospital admissions, GP records, and lifestyle habits such as smoking.
Instead of predicting the exact date of a medical event (like a heart attack on October 1st), Delphi-2M estimates the likelihood of more than 1,200 diseases, ranging from heart problems to sepsis.
🔹 Think of it like this: Just as weather apps say there’s a 70% chance of rain tomorrow, Delphi-2M can tell you there’s a 70% chance of developing a certain disease within the next few years.
Why Is This Breakthrough Important?
- Early Warnings Save Lives-Doctors could identify high-risk patients earlier, providing treatment or lifestyle advice before diseases fully develop. For example, someone at high risk of liver disorders could be encouraged to cut back on alcohol before permanent damage sets in.
- Smarter Healthcare Planning-Hospitals could use these forecasts to plan resources better. Imagine knowing in advance how many heart attack patients a city like Norwich might have by 2030. This means better staffing, facilities, and preparation.
- Personalized Medicine at Scale-Just like people today are prescribed statins if their risk of a heart attack is high, AI forecasts could expand this approach across many diseases. Personalized prevention strategies could become the norm.
Tested and Proven
The AI model was first developed using anonymous health data from the UK Biobank (over 400,000 people). It was then tested on medical records from nearly 1.9 million people in Denmark. The results? Very promising.
According to Prof. Ewan Birney of the European Molecular Biology Laboratory:
“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.”
That level of accuracy shows this is more than just an academic experiment—it’s a glimpse into the future of healthcare.
What’s Next?
Delphi-2M isn’t ready for hospitals yet, but the vision is clear:
- Spot diseases earlier.
- Tailor prevention strategies.
- Anticipate healthcare needs years in advance.
As Prof. Moritz Gerstung of the German Cancer Research Centre put it:
“This is the beginning of a new way to understand human health and disease progression.”
Just as weather forecasts transformed how we prepare for our day, AI health forecasts could transform how we prepare for our future. Instead of reacting to diseases, we could prevent them. Instead of uncertainty, we could have probabilities.
AI won’t replace doctors, but it will give them—and us—a new kind of foresight. The question is, are we ready to act on the forecast?