# Can saliency maps flag chest radiograph classifier errors at a new hospital? An external validation study

DevFeed: [Can saliency maps flag chest radiograph classifier errors at a new hospital? An external validation study](<https://devfeed.tech/articles/can-saliency-maps-flag-chest-radiograph-classifier-errors-at-a-new-hospital-an-external-validation-study-117003.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.05011>)

Author: Aashish Chapagai, Angad Singh Gurtatta, Erik Cambria

Published: 2026-11-04T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [model-deployment](<https://devfeed.tech/topics/model-deployment.md>), [Calibrated uncertainty](<https://devfeed.tech/topics/calibrated-uncertainty.md>), [machine learning overfitting](<https://devfeed.tech/topics/machine-learning-overfitting.md>)

Tags: [computer](<https://devfeed.tech/tags/computer.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [eess-iv](<https://devfeed.tech/tags/eess-iv.md>), [errors](<https://devfeed.tech/tags/errors.md>), [external](<https://devfeed.tech/tags/external.md>), [hospital](<https://devfeed.tech/tags/hospital.md>), [maps](<https://devfeed.tech/tags/maps.md>), [validation](<https://devfeed.tech/tags/validation.md>)

## AI overview

In an external validation of chest radiograph classifiers at two hospitals in the USA and Vietnam, Grad-CAM localized findings less often than radiologists (hit rate at most 0.575 versus 0.888). Adding anatomical enrichment of saliency maps to a confidence-based error detector calibrated at the training hospital improved AUC by no more than 0.011. For the convolutional network, the relationship between enrichment and correctness reversed across hospitals and degraded the transferred detector. These results suggest model confidence should anchor deployment monitoring; the supplied abstract does not establish that saliency is useful for detecting individual errors at new sites.

## Source excerpt

arXiv:2611.05011v1 Announce Type: cross Abstract: Saliency maps are offered as a safeguard for medical image classifiers, but whether they reveal individual errors at a new hospital remains largely untested. We trained a convolutional network and a vision transformer on NIH ChestX-ray14 and evaluated 303,768 decisions per model at the training hospital and at two external sites, in the USA and Vietnam, with radiologist-consensus labels. We scored each explanation by its anatomical enrichment, the area-normalized concentration of attribution in a segmented thoracic region, and tested whether it detects errors beyond the model's confidence. Grad-CAM localized findings less often than radiologists (hit rate at most 0.575 versus 0.888). Adding enrichment to a confidence-based error detector calibrated at the training hospital never raised its area under the curve by more than 0.011, and for the convolutional network the relation between enrichment and correctness reversed between hospitals, degrading the transferred detector. Model confidence, not saliency, should anchor deployment monitoring.