From algorithm performance to clinical value in AI-assisted skin cancer diagnostics.

What was studied

This scoping review asked what AI skin cancer tools have shown beyond accuracy. It searched PubMed/MEDLINE, Scopus and Web of Science. The work ran from January 2015 to October 2025. It found 4,021 records. Of these, 33 studies met the criteria. The authors sorted outcomes into clinical, process and system-level groups.

What they found

Most outcomes were about process and user experience. They included trust in the diagnosis, work speed and user acceptance. Only a few studies reported clinical outcomes. Fewer still reported system or cost outcomes. Most studies ran in controlled test settings. Real-world studies that look forward in time were less common.

What it means for patients

The authors say AI may help the diagnostic process, doctor confidence and decisions. They say evidence of clinical and system value is limited. They say new ways to judge outcomes are needed.

Why it is not certain yet

This review maps what has been studied. It does not test if AI helps patient health. Few studies measured clinical or cost outcomes. The authors say future work should add real-world studies and cost checks.

Source

From algorithm performance to clinical value in AI-assisted skin cancer diagnostics.

We restate facts from the source record in our own words. The original wording belongs to its publisher.