AI Science Tracker
EN

COVID-Net and CoroNet show weaker performance on external chest X-ray data

Emory University; Indiana University Purdue University Indianapolis; Mentor Graphics India; K. N. Toosi University of Technology · 23 Jun 2020 · medicine

Discovery

COVID-Net and CoroNet show weaker performance on external chest X-ray data

Source check: Source checked. Not independent scientific confirmation.

Finding: The v1 paper externally evaluates the pretrained COVID-Net and CoroNet chest-radiograph classifiers on three pre-pandemic public datasets and an Emory RT-PCR cohort. It finds markedly degraded generalization, including high false positives on pre-COVID images and poor COVID-Net sensitivity/precision on the institutional cohort; this is a specific retrospective diagnostic-validation result, not evidence of clinical benefit.

Technology: The authors apply two already-trained deep-learning CXR classifiers, COVID-Net and CoroNet, to independent public and institutional datasets; the models provide the COVID-19 classi

Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
The authors apply two already-trained deep-learning CXR classifiers, COVID-Net and CoroNet, to independent public and institutional datasets; the models provide the COVID-19 classifications under evaluation.
Test and result
The Results report COVID-Net false-positive rates of 55.3% on CheXpert and 23.4% on MIMIC-CXR, and on Emory 61.4% sensitivity, 0.54 F1 and 0.49 precision; CoroNet has a 93.7% false-negative rate on Emory. The central result is external generalization failure.
Limitation / still unknown
Evaluation is retrospective and uses datasets with different label mechanisms; the paper notes PCR positivity and radiographic findings are not identical. No clinical benefit was tested. arXiv metadata did not provide an affiliation, so the record states that the affiliation was not provided.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/pdf/2006.13262v1 · https://arxiv.org/abs/2006.13262v1 · https://export.arxiv.org/api/query?id_list=2006.13262 · https://arxiv.org/pdf/2006.13262v1#Materials-and-Methods · https://arxiv.org/pdf/2006.13262v1#Results

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse medicine discoveries.