# Cross-dataset transportability of pediatric chest X-ray deep learning across three countries: discrimination, calibration, operating-point failure, and limited-label recovery

arXiv: https://arxiv.org/abs/2609.05140 · HTML: https://mlsanity.com/paper/2609.05140/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 7 of its 25 references that are in the map.
- First posted on arXiv 4 Sep 2026.

## Builds on (most cited first, 7 of 7)

- [Densely Connected Convolutional Networks](https://mlsanity.com/paper/1608.06993/) (2016)
- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Domain Generalization with MixStyle](https://mlsanity.com/paper/2104.02008/) (2021)
- [TorchXRayVision: A library of chest X-ray datasets and models](https://mlsanity.com/paper/2111.00595/) (2021)
- [Deep Learning for Chest X-ray Analysis: A Survey](https://mlsanity.com/paper/2103.08700/) (2021)
- [Deep Learning Applied to Chest X-Rays: Exploiting and Preventing Shortcuts](https://mlsanity.com/paper/2009.10132/) (2020)
- [PediCXR: An open, large-scale chest radiograph dataset for interpretation of common thoracic diseases in children](https://mlsanity.com/paper/2203.10612/) (2022)

## Abstract (arXiv)

Background and Objective: External evaluation of medical-imaging AI is often collapsed into discrimination. We evaluated a computational protocol that separately tests discrimination, probability calibration, fixed operatingpoint transport, shortcut-associated signal, and limited-label recoverability for pediatric pneumonia classification across datasets from three countries. Methods: After exact-duplicate removal, 5,824 Guangzhou radiographs supported leakage-controlled source development and internal testing. A frozen three-seed DenseNet121 dual-view ensemble was evaluated zero-shot on BDCXR-3257 from Bangladesh (n = 3, 257) and an untouched harmonized VinDr-PCXR/PediCXR test cohort from Vietnam (n = 1, 077). Matched seed-42 variants tested architectural robustness. Secondary BDCXR analyses used a fixed 651-image adaptation pool and 2,606-image hold-out; 163, 326, and 651 labels represented 5%, 10%, and 20% of complete BDCXR. Results: Internal AUROC was 0.976 with 95.1% sensitivity. BDCXR and VinDr-PCXR AUROC were 0.798 and 0.742, while frozen-threshold sensitivity fell to 6.2% and 0%. Source-to-BDCXR AUROC degradation occurred for a full-image baseline (0.961 to 0.749), ungated dual-view model (0.977 to 0.766), and gated MixStyle model (0.966 to 0.789). With 163 BDCXR labels, Platt recalibration preserved AUROC while increasing held-out sensitivity to 88.3%, but specificity was 47.9% and the alert rate was 78.5%. Two hundred repeated 163-label fits confirmed sensitivity recovery but substantial specificity variability. Conclusions: Cross-dataset shifts across countries affected ranking, probability alignment, and source-defined decision behavior differently. Transport studies should evaluate these components separately and quantify the operational burden of apparent recovery.