Who decides when medical AI may be updated?

by Åsa Catapano, Karolinska Institutet

edited by Sadie Harley, reviewed by Robert Egan

Sadie Harley

Scientific Editor

Meet our editorial team
Behind our editorial process

Robert Egan

Senior Editor

Meet our editorial team
Behind our editorial process Editors' notes

This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility:

fact-checked

peer-reviewed publication

trusted source

proofread

The GIST Add to Preferred Sources

Medical AI can improve with new patient data, but every update must be checked first. Patients, equipment and ways of working change, and a model trained on older data can gradually get worse without anyone noticing.

Regulations in the U.S. and the EU state what must be done when medical AI is monitored and updated, but not in detail how. In the journal npj Digital Medicine, researchers at Karolinska Institutet present AEGIS, a framework for governing and monitoring medical AI models on the market.

Two questions at every update

AEGIS answers two questions: Can the new model be put into use? Does the model already in use need a closer look? The answers follow rules in a change plan approved before the device reached the market.

"Most AI-based medical devices are locked after approval and do not learn from new data. Every improvement needs retraining and validation," says Farhad Abtahi, head of the SMAILE core facility at the Department of Clinical Science, Intervention and Technology and corresponding author of the study.

Two demonstrations: Sepsis and brain tumors

The researchers demonstrated AEGIS with two models: one that identifies sepsis cases in past intensive care data and one that outlines brain tumors on MRI scans.

"The demonstrations examine the decision rules but do not show how safe or effective an AI model is in health care. Each clinical application requires its own validation and regulatory assessment," says Abtahi.

AEGIS currently detects changes in incoming data, but not yet directly whether the relationship between data and outcomes changes. The next step is to test the framework in real-world use with clinical partners and authorities.

Publication details

Fardin Afdideh et al., AEGIS: an operational infrastructure for post-market governance of adaptive medical AI under US and EU regulatory frameworks, npj Digital Medicine (2026). DOI: 10.1038/s41746-026-03290-1

Journal information: npj Digital Medicine

Provided by Karolinska Institutet Who's behind this story?

Sadie Harley

BSc Life Sciences & Ecology. Microbiology lab background with pharmaceutical news experience in oil, gas, and renewable industries. Full profile →

Robert Egan

Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile →

Citation: Who decides when medical AI may be updated? (2026, October 23) retrieved 23 October 2026 from https://medicalxpress.com/news/2026-10-medical-ai.html This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.