The AI That Is Helping Doctors Detect Cancer Earlier Than Ever Before

AI-assisted cancer screening is being deployed across European health systems with results that are changing what is possible in early detection. Here is the latest evidence.

The statistics are striking enough that they deserve careful presentation rather than hyperbole: in randomized controlled trials conducted across six European countries, AI-assisted mammography screening detected 20-35 percent more breast cancers than radiologist-only screening, with no increase in false positive rates. AI-assisted colonoscopy, where AI systems flag polyps in real-time during the procedure, has increased polyp detection rates by 16-24 percent in routine clinical use. AI analysis of lung CT scans has reduced the average size of lung cancers detected by screening programmes — a proxy measure for how early in development they are being found.

These are clinical results from real health systems — not from controlled research environments but from routine clinical deployment in Sweden, Denmark, the Netherlands, Germany, Italy, and the UK. They represent genuine, documented improvements in cancer detection that translate directly into improved patient survival because earlier-detected cancers are more treatable in every major cancer type that these systems are being used for.

The European Commission's Cancer Mission, which set ambitious targets for early detection in its 2021-2022 strategy documents, is now able to point to AI-assisted screening as a genuine tool for achieving those targets rather than a hoped-for future intervention.

The challenges in scaling these benefits are not primarily technical. AI diagnostic systems have been shown to work. The challenges are organizational: integrating AI systems into clinical workflows that were designed without them; managing the regulatory requirements for medical AI under the EU AI Act; ensuring equitable access across health systems with very different digital infrastructure maturity; and handling the specific medico-legal questions that arise when AI and human diagnostic conclusions differ.

The last challenge — what happens when AI and radiologist disagree about whether a scan shows cancer — is the frontier question that European health regulators are currently working through in guidance documents that will determine how much clinical authority AI systems are granted in European medical practice.