Sakhanda Wire
NVDA $154.51 -1.04% MSFT $506.49 -2.22% GOOGL $281.27 -0.86% META $621.51 -0.78% AMZN $256.90 -0.33%
← Back to the news

Bristol researchers propose drug-style checks for medical AI

Researchers at the University of Bristol have proposed a framework for assessing the reliability of medical AI before systems are used in clinical settings. Called “Learning Ensemble,” it is modeled on the structured information and testing used to determine how medicines should be used, even when their precise effects are not fully understood.

The framework asks developers to document three areas. The first covers a system’s operating limits, including its intended doctors or clinics, hardware requirements and training data. The second examines performance across patient groups rather than relying only on an overall accuracy figure. The third considers whether the system is genuinely suitable for its clinical purpose.

The researchers cite earlier examples to show why these checks matter. A 2021 COVID-19 X-ray system reportedly learned to rely on incidental image details and failed at another clinic. Another 2021 study found poorer detection of disease among underserved populations. In a separate example, an AI system classified asthma patients with pneumonia as low risk because aggressive emergency treatment had improved their outcomes, making the result unsuitable for triage. The authors describe the framework as a starting point requiring expert review and ongoing adjustment.

Written by Sakhanda Wire AI from reporting by The Decoder. Facts, quotes and full context are in the original article below.

Originally published by The Decoder on

Read the original on The Decoder ↗

The reporting and the image belong to The Decoder and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news