Document is current

Article History

Received: 25 February 2026

Revised: 20 April 2026

Accepted: 20 May 2026

First Online: 18 June 2026

Compliance with ethical standards

Guarantor: The scientific guarantor of this publication is Solveig Hofvind.

Conflict of interest: The Cancer Registry has a research agreement with Lunit Inc. for the use of their AI software. None of the authors were employees of or consultants for Lunit, which had no access to the data and were not involved in any part of the study. One author is an employee of the Norwegian Computing Center (NCC), but did not have an impact on the study sample. The remaining authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the article.

Statistics and biometry: One of the authors has significant statistical expertise (M.B.B.).

Informed consent: The data were disclosed with legal basis in the Personal Health Data Filing system Act Section 19a-h [ ] and the Cancer Registry Regulation Section 3–1 [ ].

Ethical approval: This study was approved by the Regional Committees for Medical and Health Research Ethics (#13294, #11022).

Study subjects or cohorts overlap: The study population of 130,031 screening examinations performed in Østfold as part of BreastScreen Norway, 2008–2018, overlaps with two other studies. The first study was published in Radiology: Artificial Intelligence (Martiniussen et al [ ]) and explored the overall performance of the same two AI models as used in the current study (Lunit INSIGHT MMG version 1.1.7.2 and an in-house model from the Norwegian Computing Center and the Cancer Registry of Norway). The second study was submitted to European Radiology on December 11, 2025. No analysis overlapped.

Methodology: