Concern 625 · Working draft · editorial review in progress
Radiology model drift
Radiology model drift occurs when changes in scanners, image processing or patient populations make a previously tested image-reading system less reliable.
Understanding the concern
A scan is shaped by the machine and settings used to produce it. If a hospital changes equipment or software, images may look subtly different to an AI system even when a clinician can still read them. The system's usual warning threshold may then miss more disease or flag more healthy scans.
Research on image acquisition changes shows why approval and testing cannot be a one-time event. A useful analogy is a thermometer that needs checking after its surroundings change. Hospitals should track performance after upgrades, investigate changes in error patterns and verify any adjustments using appropriate clinical evidence.
An example
A breast-screening service upgrades its image-processing software and finds that an unchanged AI threshold now produces more false alarms.
Illustrative unless identified as a documented case.
Words in plain English
- AI
- Software that learns patterns from examples or uses rules to perform tasks such as recognizing images or generating text. It can make mistakes and does not understand every situation as a person would.
- Model
- The learned part of an AI system. It uses patterns found during training to produce an answer or prediction.
- Drift
- A change over time in data or real-world conditions that can make a system less reliable.
Evidence and its limits
The paper studies mammography and pathology acquisition shifts and proposes a correction method. It does not show that every equipment change causes harm or that correction can safely be left unchecked.
Nature Communications 2023 · Model evaluation
Automatic correction of performance drift under acquisition shift in medical image classification ↗ (opens in a new tab)Mammography and pathology acquisition shifts and a proposed recalibration method. Not a guarantee of safe automatic correction in every hospital.
Broader frameworks and background
These resources provide context. They do not establish every claim in this concern.
A question to discuss
Who checks the AI again when the equipment or image-processing software changes?
Keep track of different views, what needs checking, and any next step in your own discussion notes.
Open discussion notes →Principles for the conversation
Before you finish · reflect together
Virtues and principles to consider
Virtues are qualities we practice in how we treat people and make choices. They are not scores for judging someone. Use these ideas to explore what a thoughtful response could look like.
Service
Service means using abilities and resources to meet real needs. It asks whether the people meant to benefit actually receive help, rather than treating them as a means to another goal.
Speak with the people the project is meant to serve. Notice burdens it may add, such as extra checking or confusing forms. Judge success by their experience as well as the organization's targets.
Ask together: Who is this meant to serve, and do they experience it as helpful?
Dignity
Dignity means recognizing each person's worth regardless of usefulness, status, or a system's score. People deserve respect and a voice in matters that affect them.
Explain important decisions, avoid humiliating treatment, and keep room for a person's story. Do not treat a prediction as a complete account of who someone is.
Ask together: Does this use treat each person as someone whose experience matters?
Listening
Listening means trying to understand someone's experience before preparing an answer. People who use a service, work with it, or are judged by it may notice problems its designers miss.
Invite people to describe what happened in their own words. Repeat back what you understood and let them correct it. A quiet person or a complaint that is hard to measure may still reveal something important.
Ask together: Whose experience have we not heard yet?
Editorial ratings
These are editorial judgments from the register, not measured probabilities or predictions about your situation.
About the register and scoring →More ways to explore
Search for more writing
For more specific or recent material, open a search below or ask for a web summary. AI-generated summaries can contain mistakes; open the original articles to check them.