A new study from the University of Gothenburg shows that artificial intelligence can flag individuals at high risk of developing melanoma years before a clinical diagnosis by mining existing health records. The finding points to a shift from reactive detection toward proactive, data-driven screening strategies at population scale.
Leveraging Health Data for Predictive Insights
Researchers analyzed clinical and administrative data for about 6 million Swedish adults. Models incorporated information beyond age and sex, including medication history, prior diagnoses and sociodemographic variables drawn from routine electronic health records.
The most advanced model reached roughly 73 percent accuracy in identifying future melanoma patients, compared with about 64 percent when using only age and gender. Importantly, the algorithm isolated small groups with very elevated risk, including cohorts with approximately a one in three chance of a melanoma diagnosis within five years.
“We can use data already in patient records to identify who should be followed more closely,” said Martin Gillstedt, noting the value of strategic use of available healthcare data.
Transforming Early Detection and Patient Care
Early detection matters for melanoma because outcomes worsen after the disease spreads. An AI-driven risk score can help clinicians prioritize follow-up and direct targeted screening toward those most likely to benefit, reducing unnecessary procedures for lower-risk patients.
“By concentrating monitoring on smaller high-risk groups we can use limited clinical resources more effectively,” said Sam Polesie, highlighting potential gains in efficiency and patient benefit.
The Future of AI in Diagnostics
This work demonstrates the potential of personalized risk assessments to inform future screening protocols. Before routine deployment, the models need prospective validation, assessment of equity and bias, regulatory review and practical integration into electronic health workflows with clinician training.
As researchers refine performance and policymakers set standards, this study offers a concrete example of how AI can extend beyond diagnostic support to enable proactive cancer prevention at a population level.
