Scientists looked at health data from every adult in Sweden. They used information like age, gender, medical history, medicines, and where people live. The study covered over six million people, and during five years, about 38,500 of them got melanoma, a type of skin cancer.
What the researchers found
Martin Gillstedt, a PhD student, said the data already in hospitals can help find people who might get melanoma later. This is not something doctors use today, but the study shows it could be possible.
AI makes predictions better
The team tried several computer models that learn from data. The best model guessed correctly about 73% of the time whether a person would develop melanoma. Using only age and gender gave about 64% accuracy.
When the models included medical diagnoses, medicines, and social information, they could pick out small groups with a much higher risk. In those groups, about one in three people got melanoma within five years.
Screening the right people could save resources
Sam Polesie, a dermatologist, explained that checking only the high‑risk groups could make screening more accurate and cheaper. It would mix big‑picture data with what doctors see in the clinic.
Looking ahead
Even though the results are promising, more research and policy work are needed before this method can become routine. The study shows that AI trained on large health registries could help create personal risk scores and improve future melanoma‑screening plans.
The project was done together by the University of Gothenburg and Chalmers University of Technology.