A landmark Swedish study of over six million adults shows that artificial intelligence, trained on routine health records, can identify people at risk of skin cancer with 73% accuracy, years before any symptom appears.
Catching cancer before it starts
Melanoma is among the most dangerous cancers precisely because of its speed. Once it penetrates through the skin and reaches the bloodstream, survival rates fall sharply. The best defence has always been early detection, ideally before a suspicious mole has had any chance to evolve into something life-threatening.
Now, a landmark study from Sweden suggests that artificial intelligence can identify the people most at risk up to five years before a diagnosis is made, using only health data that medical systems already routinely collect.
According to research published in Acta Dermato-Venereologica, a team from the University of Gothenburg and Chalmers University of Technology, analysed national registry data covering the entire adult population of Sweden.
With over six million individuals included and 38,582 subsequently developing melanoma over a five-year observation window, this is among the largest datasets ever assembled for cancer risk modelling. The results point toward a fundamental shift in how melanoma screening could work, from broad, reactive checks to precise, AI-guided prevention.
How the study worked
The research did not rely on specialist dermatology records, genetic sequencing, or expensive imaging. According to EurekAlert, the AI models were fed entirely routine administrative health data: age, sex, medical diagnoses, medication use, and socioeconomic status, information that healthcare systems in most developed countries capture as a matter of standard record-keeping.
Researchers then asked whether machine learning could detect patterns in that unremarkable-seeming data that predicted who would develop melanoma within the following five years. The answer was an unambiguous yes.
The 33% risk threshold that changes everything
The nine-percentage-point accuracy improvement between a basic model and the advanced AI may sound modest, but across a population of millions it translates into thousands of additional melanomas caught early enough to treat effectively. More striking still was the AI’s ability to isolate very small sub-groups with dramatically elevated risk.
According to the published study results cited by Euronews Health, when the model combined diagnoses, medication histories, and sociodemographic data, it pinpointed groups where the likelihood of developing melanoma within five years reached approximately 33 percent.
For context, the baseline population risk was just 0.64 percent. That 33-percent figure is roughly 50 times the average, a level of precision that transforms a screening tool from theoretically useful to clinically actionable.
“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. This would involve bringing population data into precision medicine and supplementing clinical assessments.” Sam Polesie, Associate Professor of Dermatology, University of Gothenburg, lead author.
Read More: Meet MangroveGS, the AI Tool that Predicts Cancer with accuracy
Why early melanoma detection matters so much
Melanoma accounts for approximately four percent of all new cancer diagnoses in the European Union, according to the European Commission’s Joint Research Centre. Despite representing a small fraction of all skin cancers by volume, it is responsible for the vast majority of skin cancer deaths because of its capacity to metastasise rapidly to other organs.
When detected at Stage I, melanoma is surgically removable and long-term survival rates are high. By Stage IV, after spread to other organs, survival rates fall dramatically. Every shift toward earlier detection therefore carries an outsized impact on mortality. According to a review published in PMC on AI and melanoma diagnosis, early identification before symptoms develop remains the single most effective intervention available.
Most AI melanoma tools analyse dermoscopy images, photographs of skin lesions. This study is different: it predicts risk before any lesion exists, using invisible patterns in a patient’s medical history. No camera. No dermatologist visit. No symptom required.
What patterns does the AI actually detect?
A natural question is what hidden signals the model identified in routine health records. While AI models of this kind involve complex pattern recognition that resists simple interpretation, the inclusion of medication history as a predictor is particularly telling. Certain drugs, including immunosuppressants prescribed to organ transplant recipients, are known to elevate skin cancer risk as a side effect. Some autoimmune medications carry similar risk profiles.
Socioeconomic status correlates with sun exposure habits, outdoor occupational history, and access to preventive dermatology care. Prior diagnoses involving immune dysregulation may flag individuals whose skin is more susceptible to DNA damage from ultraviolet radiation.
According to Martin Gillstedt, a doctoral student at the University of Gothenburg’s Sahlgrenska Academy, “data which is already available within healthcare systems can be used to identify individuals at higher risk of melanoma” as reported by ScienceDaily.
From population data to precision medicine
The study illustrates a broader transformation underway in healthcare: the shift from treating patients after they present with symptoms to identifying vulnerability before disease develops. Rather than inviting every adult over a certain age for skin checks, a strategy generating enormous cost and modest incremental benefit, health systems could use AI risk scores to direct limited dermatology capacity precisely where it is most needed.
According to researchers at ICT&Health, this study “marks the next step in the development of data-driven, predictive healthcare” pointing toward a future where an AI flag in your health record might prompt a targeted invitation for screening years before you notice anything unusual on your skin.
“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.” Martin Gillstedt, doctoral student, University of Gothenburg Sahlgrenska Academy.
Read More: Scientists Create “Tumor-Eating” Bacteria That Can Destroy Cancer From the Inside
Significant barriers remain before clinical use
The researchers are explicit that their findings require further validation across different populations and healthcare systems, along with policy decisions around data privacy, ethical frameworks, equitable access, and workflow integration before AI risk scores could become a routine part of medical records.
There are also critical questions about what patients and clinicians would do with a high-risk designation. Melanoma has no approved chemoprevention drug. The primary intervention for flagged individuals is heightened vigilance, more frequent skin examinations, UV protection counselling, and dermatology referrals, all of which require patient engagement and available healthcare capacity.
Equity concerns are equally important. According to researchers and commentators cited in coverage by NewKerala, whether AI-powered screening narrows or widens the gap between urban and rural populations, and between different socioeconomic groups, will be a defining question as the technology approaches clinical implementation.
The bigger picture: AI rewriting cancer prevention
The Swedish melanoma study is part of a broader wave of research demonstrating that large-scale AI models, trained on population health records, can surface cancer risk signals invisible to conventional clinical methods. Similar approaches are being explored for bowel cancer, lung cancer, and cardiovascular disease, where the goal is the same: identifying high-risk individuals early, when intervention is most effective and least costly.
For melanoma specifically, the combination of a visually identifiable disease, a clear intervention pathway (early surgical removal), and the availability of population-wide registry data in countries like Sweden makes it an ideal testbed for this kind of predictive AI.
If the model validates across international populations, including those with different skin types, sun exposure patterns, and healthcare system structures, it could serve as a template for AI-assisted early cancer detection globally.
For more science news and research coverage, visit the Science section at bdesk.news.
Michaela Reeds is an investigative journalist and reporter with a focus on politics, science, and technology. She brings clarity to complex issues, translating policy developments, scientific breakthroughs, and technological innovations into compelling stories for a broad audience. She is known for her dedication to accuracy, transparency, and in‑depth reporting.
