Article

The NLP-to-Expert Gap in Chest X-ray AI

2026-03-02

Abstract excerpt

In previous work, we achieved state-of-the-art performance on ChestX-ray14 (ROC-AUC 0.940, F1 0.821) using pretraining diversity and clinical metric optimization. Applying the same methodology to CheXpert, we received similar results when using NLP valuation and test data—but when evaluated against expert radiologist labels, performance was only 0.75-0.87 ROC-AUC. The models had learned to match the automated NLP...

Topics

Open a Topic to create a Post that cites this publication.

Identifiers and source

Literature Corpus work: 6b2ada84-3460-5104-ae22-1bbc03e9ed80
DOI: 10.64898/2026.02.27.26347261

Open publication

Related research

Semantic proximity does not establish scientific evidence.

Click a neighbor to travelStep 1 · 12 closest

The NLP-to-Expert Gap in Chest X-ray AIDOI 10.64898/2026.02.27.26347261Read

Pretraining Diversity and Clinical Metric Optimization Achieve State-of-the-Art Performance on ChestX-ray14Semantic proximity Read

AI-Assisted Pneumonia Detection, Localisation and Report Generation from Chest X-raysSemantic proximity Read

Vision-Language Foundation Models Do Not Transfer to Medical Imaging Classification: A Negative Result on Chest X-ray DiagnosisSemantic proximity Read

Developing Deep Learning Models for Classifying Medical Imaging DataSemantic proximity Read

Cethraian-Shift is a reproducible medical-imaging research project evaluating the cross-dataset reliability, calibration, label-policy sensitivity, and Grad-CAM++ stability of multi-label chest X-ray classifiers across NIH ChestX-ray14 and VinDr-CXRSemantic proximity Read

Self-Supervised Pretraining Enables High-Performance Chest X-Ray Interpretation Across Clinical DistributionsSemantic proximity Read

High Consistency, Limited Accuracy: Evaluating Large Language Models for Binary Medical DiagnosisSemantic proximity Read

IHRAS: Automated Medical Report Generation from Chest X-rays via Classification, Segmentation, and LLMsSemantic proximity Read

Pillar-0: A New Frontier for Radiology Foundation ModelsSemantic proximity Read

The Challenge Dataset – simple evaluation for safe, transparent healthcare AI deploymentSemantic proximity Read

Algorithmic encoding of protected characteristics in chest X-ray disease detection modelsSemantic proximity Read

Large Language Models for Classifying Usual Interstitial Pneumonia from Radiology Reports: Native Reasoning Versus Structured PromptingSemantic proximity Read