Swedish researchers have used AI to spot high-risk melanoma patients before they develop the disease, using health data from existing medical records.
Artificial intelligence can identify those at higher risk of developing melanoma within five years, a new study has found.
ADVERTISEMENT
ADVERTISEMENT
Using AI, researchers from the University of Gothenburg in Sweden have identified small groups within the population that have a significantly higher risk of developing the serious, often fast-spreading form of skin cancer over the next five years.
“Our study shows that data which is already available within healthcare systems can be used to identify individuals at higher risk of melanoma,” said Martin Gillstedt, a doctoral student at the University of Gothenburg.
“This is not a form of decision support that is currently available in routine healthcare, but our results give a clear signal that registry data can be used more strategically in the future," Gillstedt explained.
The research team analysed clinical data from all Swedish adults who had been living in the country between 2005 and 2014, or some 6 million people.
They fed AI models clinical information beyond the usual age and gender, including medication history and other diagnoses.
The most advanced AI model could distinguish individuals who subsequently developed melanoma from those who did not in about 73% of cases, compared with about 64% when only age and gender were used.
By combining diagnoses, medication, and sociodemographic data, the researchers were also able to pinpoint small, high-risk groups for whom the risk of developing melanoma within five years was around 33%, the authors noted.
Early detection key
The main cause of melanoma is ultraviolet light, which naturally comes from the sun and is artificially produced in sunbeds.
It can spread to other parts of the body, and once it has spread, the survival rate drops significantly, which is why early detection is crucial.
It is estimated that skin melanoma accounted for 4% of all new cancer diagnoses in the EU in 2020 and for 1.3% of all deaths due to cancer.
This made it the sixth most frequently occurring cancer and one of the 20 most frequent causes of cancer death, according to the European Commission’s Joint Research Centre.
By identifying people with higher predicted risk, clinicians can prioritise follow-up and invite these individuals via mail or digital outreach for screening appointments, the authors wrote.
“Our analyses suggest that selective screening of small, high-risk groups could lead to both more accurate monitoring and more efficient use of healthcare resources,” lead author of the study Sam Polesie said.
The results show that AI models trained on large amounts of registry data could become key for more personalised risk assessments and future screening strategies.
However, the researchers emphasise that more research and policy decisions are needed before the method can be introduced into routine healthcare.
