# Two medical AI models missed every collapsed lung on chest X-rays

An audit of five open medical image-reading models finds that a single accuracy score can hide total failure on one diagnosis. Two models never flagged a pneumothorax, yet explained their wrong answers fluently.

- Category: Safety & evaluation
- Readiness level: II (Benchmarked) on Science Underway's four-step scale, as proposed
- Published: 2026-11-15
- Editor: Stefan Perdal
- URL: https://scienceunderway.com/articles/two-medical-ai-models-missed-every-collapsed-lung-on-chest-x-rays

## The study

Aggregate Accuracy Hides Diagnosis-Specific Failure in Medical Vision-Language Models (preprint, not peer reviewed). arXiv 2611.14975. https://arxiv.org/abs/2611.14975

## At a glance

- What it shows: Two of five open AI models that read chest X-rays did not spot a single collapsed lung across hundreds of cases. Two other models caught many of the same cases, and an overall accuracy score did not reveal the gap.
- What it does not show: It does not show how these models behave in real hospitals, or why the failure happens. It was tested on yes/no questions about single images.
- Why it matters: Buyers and regulators often compare medical AI by one average score. This study shows that score can hide a diagnosis the model never makes, delivered with a confident, plausible explanation.

## What the study shows

The authors tested five open medical vision-language models. These are AI systems that look at an image and answer in text. Each model answered 6,048 yes/no questions about chest X-rays from a public dataset called ChestX-ray14. Each of seven findings had 432 X-rays where it was present and 432 where it was absent. The paper is a preprint accepted at a NeurIPS 2026 workshop and has not been peer reviewed.

The headline result concerns pneumothorax, a collapsed lung that can kill within hours if untreated. Two models, HuatuoGPT-Vision-7B and Qwen2.5-VL-7B, answered "no" to every pneumothorax question. On the same images, Lingshu-7B found about two thirds of cases and MedGemma-4B found 38%. The failing models still wrote specific explanations, describing the lungs as fully inflated, with no hedging.

Overall accuracy did not show this. MedGemma scored highest overall, about 70% against a coin-flip level of 50%. Yet it had its own problem: it caught 95% of atelectasis, a partly collapsed lung, but wrongly flagged about two thirds of healthy cases.

The authors report the pneumothorax failure held across four ways of asking the question. It held at full image resolution and on X-rays showing only pneumothorax. It also repeated on a second dataset, SIIM-ACR, where radiologists had outlined each collapsed lung by hand. There, the failing models found none of 432 cases, including none of the 144 largest.

A tenfold larger Qwen model, at 72 billion parameters, caught only about 4% of pneumothorax cases. MedGemma did improve at its larger size, so scale helped one model family but not the other.

The authors also looked inside the models. The failing models were highly confident in their wrong "no" answers. Yet a simple classifier trained on their internal signals, called a linear probe, could partly tell pneumothorax cases apart. The authors read this as the finding being registered inside the model but lost before the answer.

## Our reading

In our reading, the strongest part of this paper is the evaluation design, not the mechanism. Balancing every finding, age band and sex means a model cannot look good by leaning on a common diagnosis. That design made the collapse visible. The second dataset with radiologist outlines answers the obvious worry that the original labels were wrong.

The practical lesson is simple. A model can score respectably on average and never make one diagnosis at all. Worse, its explanation reads like a careful negative. Imagine a triage tool built on Qwen2.5-VL-7B sorting overnight X-rays. A collapsed lung would be marked "no" with a tidy rationale, and nothing in the output would prompt a second look.

We rate this work at level II, Benchmarked. It reports results on public datasets for named open models. It does not reach level III, because the authors have not released their code or their list of test items.

The rest of the article (what it means, limits, what to watch) is free for members: https://scienceunderway.com/articles/two-medical-ai-models-missed-every-collapsed-lung-on-chest-x-rays

## How this was made

An AI model read the study and wrote this article in the structure Science Underway uses. An editor reviewed it and approved it for publication. Check the original study before you rely on the details.

---
Source: Science Underway, https://scienceunderway.com/. Please link to the article when you cite it.
