Disease Classification2019en
CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels
224,316 chest radiographs from 65,240 patients with 14 pathology labels. Includes uncertainty labels and expert radiologist annotations for validation set. The gold standard for chest X-ray classification.
Current State of the Art
CheXpert AUC Maximizer
Stanford
93
auroc
auroc Progress Over Time
Showing 4 breakthroughs from Jan 2019 to Dec 2025
Key Milestones
Jan 2019
DenseNet-121 (Chest X-ray)
Baseline DenseNet-121. Trained on CheXpert training set.
86.5
Dec 2025
CheXpert AUC MaximizerCurrent SOTA
Mean AUC across 5 competition pathologies. Competition-winning ensemble.
93.0
+4.4%
Total Improvement
7.5%
Time Span
7y 1m
Breakthroughs
4
Current SOTA
93.0
Top Models Performance Comparison
Top 7 models ranked by auroc
Best Score
93.0
Top Model
CheXpert AUC Maxi...
Models Compared
7
Score Range
6.5
aurocPrimary
| # | Model | Score | Paper / Code | Date |
|---|---|---|---|---|
| 1 | CheXpert AUC MaximizerOpen Source Stanford | 93 | Dec 2025 | |
| 2 | BioViLOpen Source Microsoft | 89.1 | Dec 2025 | |
| 3 | CheXzeroOpen Source Harvard/MIT | 88.6 | Dec 2025 | |
| 4 | GLoRIAOpen Source Stanford | 88.2 | Dec 2025 | |
| 5 | MedCLIPOpen Source Research | 87.8 | Dec 2025 | |
| 6 | TorchXRayVisionOpen Source Cohen Lab | 87.4 | Dec 2025 | |
| 7 | DenseNet-121 (Chest X-ray)Open Source Research | 86.5 | Dec 2025 |