List your company
ResearchFormerly WizWand

Disease Classification on CheXpert

0.9238AUROCCARZero
0.5048880.6136440.72240.831156Nov 18, 2025Dec 20, 2025Jan 22, 2026Feb 24, 2026Mar 28, 2026Apr 30, 2026Jun 2, 2026
Updated 5mo ago

Evaluation results

MethodLinks
CARZeroZero-shot=true2026.050.923840.286.93
KEPILZero-shot=true2026.050.912165.5687.88
X-WINPretrain. data=372K CXRs + 32K CT volumes, Backbone=ViT-Large, Evaluation protocol=linear probing2025.110.908——
X-WINPretrain. data=372K CXRs + 32K CT volumes, Backbone=ViT-Base, Evaluation protocol=linear probing2025.110.902——
MAVLZero-shot=true2026.050.901365.4786.44
DeViDeZero-shot=true2026.050.898762.8887.98
MGCADirect Inference=Yes, Labeling ratio=100%2026.060.897——
KADZero-shot=true2026.050.892363.2586.25
Medical MAEDirect Inference=No, Labeling ratio=100%2026.060.892——
IMT-CXRDirect Inference=Yes, Labeling ratio=100%2026.060.892——
MGCADirect Inference=Yes, Labeling ratio=10%2026.060.891——
IMT-CXRDirect Inference=Yes, Labeling ratio=10%2026.060.89——
MGCADirect Inference=Yes, Labeling ratio=1%2026.060.888——
MRMDirect Inference=No, Labeling ratio=100%2026.060.887——
MRMDirect Inference=No, Labeling ratio=1%2026.060.885——
MRMDirect Inference=No, Labeling ratio=10%2026.060.885——
IMT-CXRDirect Inference=Yes, Labeling ratio=1%2026.060.885——
REFERSDirect Inference=No, Labeling ratio=100%2026.060.882——
REFERSDirect Inference=No, Labeling ratio=10%2026.060.881——
GloRIADirect Inference=Yes, Labeling ratio=100%2026.060.881——
MedKLIPZero-shot=true2026.050.879763.6784.32
CheXzeroZero-shot=true2026.050.87961.981.17
M3AEDirect Inference=No, Labeling ratio=100%2026.060.879——
GloRIADirect Inference=Yes, Labeling ratio=10%2026.060.878——
CheXFoundPretrain. data=987K CXRs, Backbone=ViT-Large, Evaluation protocol=linear probing2025.110.876——
Ark+Pretrain. data=704K CXRs, Backbone=Swin-Base, Evaluation protocol=linear probing2025.110.876——
M3AEDirect Inference=No, Labeling ratio=10%2026.060.873——
ConVIRTDirect Inference=Yes, Labeling ratio=100%2026.060.873——
REFERSDirect Inference=No, Labeling ratio=1%2026.060.872——
CheXWorldPretrain. data=448K CXRs, Backbone=ViT-Base, Evaluation protocol=linear probing2025.110.871——
ConVIRTDirect Inference=Yes, Labeling ratio=10%2026.060.868——
GloRIADirect Inference=Yes, Labeling ratio=1%2026.060.866——
MaCoPretrain. data=377K CXRs w/ reports, Backbone=ViT-Base, Evaluation protocol=linear probing2025.110.864——
M3AEDirect Inference=No, Labeling ratio=1%2026.060.862——
ConVIRTDirect Inference=Yes, Labeling ratio=1%2026.060.859——
CARZeroPretrain. data=377K CXRs w/ reports, Backbone=ViT-Base, Evaluation protocol=linear probing2025.110.852——
RAD-DINOPretrain. data=LVD-142M + 838K CXRs, Backbone=ViT-Base, Evaluation protocol=linear probing2025.110.846——
CheXAgentPretrain. data=1.07M CXRs w/ text, Backbone=ViT-Large, Evaluation protocol=linear probing2025.110.813——
CXR-AlignPretrain. data=325K CXRs w/ reports, Backbone=ViT-Base, Evaluation protocol=linear probing2025.110.793——
DINOv2Pretrain. data=LVD-142M, Backbone=ViT-Base, Evaluation protocol=linear probing2025.110.776——
I-JEPAPretrain. data=ImageNet-1K, Backbone=ViT-Large, Evaluation protocol=linear probing2025.110.764——
BioViL-TZero-shot=true2026.050.709347.2169.96
BioViLZero-shot=true2026.050.600142.166.13
GLoRIAZero-shot=true2026.050.548437.8660.7
ConVIRTZero-shot=true2026.050.52135.6157.43