We use the open-source mimic-cxr repository4 to extract impression and findings for each report. Following [9], we pick out sequences of alphanumeric characters and drop all other characters and symbols for all reports, and remove reports which contain less than3 tokens. Following common practice in ViT [5], we split the radiograph with patch size16 16,which results in 196 visual tokens for each image. The instance-level projection layer is a two-layer MultiLayer Perceptron (MLP) with Batch Normalization [10] and ReLU activation function. Additionally, we use a frozen Batch Normalization layer after the MLP toobtain instance-levelembeddings.

Artificial intelligence is revealing what medicine never intended to encode. Deep vision models, trained on chest X-rays, can now detect not only disease but also invisible traces of social inequality. In this study, we show that state-of-the-art architectures (DenseNet121, SwinV2-B, MedMamba) can predict a patient's health insurance type, a strong proxy for socioeconomic status, from normal chest X-rays with significant accuracy (AUC around 0.70 on MIMIC-CXR-JPG, 0.68 on CheXpert). The signal was unlikely contributed by demographic features by our machine learning study combining age, race, and sex labels to predict health insurance types; it also remains detectable when the model is trained exclusively on a single racial group. Patch-based occlusion reveals that the signal is diffuse rather than localized, embedded in the upper and mid-thoracic regions. This suggests that deep networks may be internalizing subtle traces of clinical environments, equipment differences, or care pathways; learning socioeconomic segregation itself. These findings challenge the assumption that medical images are neutral biological data. By uncovering how models perceive and exploit these hidden social signatures, this work reframes fairness in medical AI: the goal is no longer only to balance datasets or adjust thresholds, but to interrogate and disentangle the social fingerprints embedded in clinical data itself.

Building robust medical machine learning systems requires pretraining strategies that exploit the intrinsic structure present in clinical data. We introduce Multiview Masked Autoencoder (MVMAE), a self-supervised framework that leverages the natural multi-view organization of radiology studies to learn view-invariant and disease-relevant representations. MVMAE combines masked image reconstruction with cross-view alignment, transforming clinical redundancy across projections into a powerful self-supervisory signal. We further extend this approach with MVMAE-V2T, which incorporates radiology reports as an auxiliary text-based learning signal to enhance semantic grounding while preserving fully vision-based inference. Evaluated on a downstream disease classification task on three large-scale public datasets, MIMIC-CXR, CheXpert, and PadChest, MVMAE consistently outperforms supervised and vision-language baselines. Furthermore, MVMAE-V2T provides additional gains, particularly in low-label regimes where structured textual supervision is most beneficial. Together, these results establish the importance of structural and textual supervision as complementary paths toward scalable, clinically grounded medical foundation models.

We visualize the domain adaptation performance on the transfer scenario NIH-CXR14, CheXpert, MIMIC-CXR to Open-i . The visualization sample in the Open-i is suffering from Atelecsis and Effusion disease.

Work done as a master's student at the University of Chicago. One way to compute this optimum is to use EM. We then get the following (see e.g., Sec 4.2.4 of Murphy [2022] In this section we discuss the datasets in more detail. B.1 Colored MNIST We show some sample images in Figure 1. We show some sample images in Figure 2. We list all the target attributes in Table 1.

Medical imaging foundation models must adapt over time, yet full retraining is often blocked by privacy constraints and cost. We present a continual learning framework that avoids storing patient exemplars by pairing class conditional diffusion replay with Elastic Weight Consolidation. Using a compact Vision Transformer backbone, we evaluate across eight MedMNIST v2 tasks and CheXpert. On CheXpert our approach attains 0.851 AUROC, reduces forgetting by more than 30\% relative to DER\texttt{++}, and approaches joint training at 0.869 AUROC, while remaining efficient and privacy preserving. Analyses connect forgetting to two measurable factors: fidelity of replay and Fisher weighted parameter drift, highlighting the complementary roles of replay diffusion and synaptic stability. The results indicate a practical route for scalable, privacy aware continual adaptation of clinical imaging models.

Artificial intelligence has shown significant promise in chest radiography, where deep learning models can approach radiologist-level diagnostic performance. Progress has been accelerated by large public datasets such as MIMIC-CXR, ChestX-ray14, PadChest, and CheXpert, which provide hundreds of thousands of labelled images with pathology annotations. However, these datasets also present important limitations. Automated label extraction from radiology reports introduces errors, particularly in handling uncertainty and negation, and radiologist review frequently disagrees with assigned labels. In addition, domain shift and population bias restrict model generalisability, while evaluation practices often overlook clinically meaningful measures. We conduct a systematic analysis of these challenges, focusing on label quality, dataset bias, and domain shift. Our cross-dataset domain shift evaluation across multiple model architectures revealed substantial external performance degradation, with pronounced reductions in AUPRC and F1 scores relative to internal testing. To assess dataset bias, we trained a source-classification model that distinguished datasets with near-perfect accuracy, and performed subgroup analyses showing reduced performance for minority age and sex groups. Finally, expert review by two board-certified radiologists identified significant disagreement with public dataset labels. Our findings highlight important clinical weaknesses of current benchmarks and emphasise the need for clinician-validated datasets and fairer evaluation frameworks.

No one doubts that our future will feature more automation than our past or present. The question is how we get from here to there, and how we do so in a way that is good for humanity. Sometimes it seems the most direct route is to automate wherever possible, and to keep iterating until we get it right. Here's why that would be a mistake: imperfect automation is not a first step toward perfect automation, anymore than jumping halfway across a canyon is a first step toward jumping the full distance. Recognizing that the rim is out of reach, we may find better alternatives to leaping--for example, building a bridge, hiking the trail, or driving around the perimeter. This is exactly where we are with artificial intelligence. AI is not yet ready to jump the canyon, and it probably won't be in a meaningful sense for most of the next decade. Rather than asking AI to hurl itself over the abyss while hoping for the best, we should instead use AI's extraordinary and improving capabilities to build bridges.