OmniMed‑FL is a multimodal federated learning framework that fuses chest radiographs and synthetic patient notes to classify five clinical conditions. The study benchmarks eight fusion strategies, three initializations, and four missing‑text imputation rules across 3–20 hospital clients under non‑IID Dirichlet partitioning, showing that federated approaches (FedAvg, FedProx, SCAFFOLD‑AdamW) outperform local‑only training. Multimodal fusion consistently improves performance, achieving macro‑F1 scores up to 0.956 on the synthetic corpus and 0.906 on the radiograph corpus.
By Ayush Debnath, Ruelia Saha, Sudip Misra
The paper evaluates federated learning with Low‑Rank Adaptation (LoRA) for fine‑tuning the BiomedCLIP vision‑language model on chest X‑ray classification across four international cohorts. Federated LoRA improves shared‑class AUC from 0.687 to 0.802, outperforming isolated single‑cohort training and approaching a centralized reference. The study shows that SVD‑based product‑space aggregation (FlexLoRA) is crucial for performance, while FedProx offers no advantage over FedAvg in this setting.
By Sanjaya Poudel, Nirajan Kunwor, Manish Dhakal, Debesh Jha, Sunil Kumar Gaire
arXiv:2509. 10517v3 Announce Type: replace Abstract: Machine learning can predict in-hospital mortality, but data privacy and the statistical heterogeneity of clinical data hamper its use.
By Rodrigo Tertulino
arXiv:2607. 20641v1 Announce Type: new Abstract: Federated learning (FL) enables multiple clinical institutions to collaboratively train a shared disease classifier without centralizing patient data.
By Afsaneh Mahanipour, Hana Khamfroush
GAMF-Net is a grouped availability‑aware multimodal fusion network designed for smartphone skin lesion classification. It divides ten clinical variables into three semantic groups—demographics, history‑related risk, and lesion presentation—and encodes each group with field values and availability masks. The image branch supplies base logits while each clinical group contributes image‑conditioned residual evidence; only available groups are fused with equal weight, and the model defaults to image‑only when no clinical data are present. Evaluated on PAD‑UFES‑20, GAMF-Net achieves an accuracy of 74.17 % and outperforms image‑only and other multimodal baselines, with ablation studies highlighting the importance of clinical information and constrained equal‑weight fusion.
By Yue Zhuo, Jingjia Yan, Fangyu Liu, Lingjun Liu, Zongjun Chen, Zhisheng Luo, Zhonghua Xie
FedDRAW introduces a new server‑side aggregation technique for federated learning called Federated Dual Reputation Annealing Weighting. It blends a data‑size prior with cosine similarity between client and global model parameters, using two annealing schedules to shift weight from size to similarity and to gradually relax selectivity. Experiments on 12 simulated client partitions of CheXpert and ChestMNIST datasets show FedDRAW outperforms seven federated baselines in AUC and geometric mean of sensitivity and specificity, with statistically significant results.
By Maryam Moradpour, Anne-Christin Hauschild
Fed-ReMasker is a federated learning approach that adapts the ReMasker masked autoencoder for tabular data imputation, specifically addressing feature-level missingness where entire features are absent at some centers. The method enables centers to impute unobserved features by leveraging knowledge from collaborating institutions. In benchmark tests on synthetic and real-world datasets, Fed-ReMasker achieves the lowest imputation error in the majority of scenarios and remains robust to client heterogeneity, closely matching the performance of a centralized model.
By Ioannis Papathanail, Rooholla Poursoleymani, Lubnaa Abdur Rahman, Stavroula Georgia Mougiakakou
The study evaluates how well different medical foundation models agree on image embeddings by measuring mutual k‑nearest‑neighbor overlap across 750,000 images from 14 datasets and 18 public models. Results show that label‑supervised models converge the least, sharing only 6–10% of nearest neighbors, whereas models trained with self‑distillation, masked image modeling, or contrastive pretraining share up to 80%. The authors suggest that developers can select pretraining objectives to control convergence and recommend testing unlabeled images for nearest‑neighbor consistency before replacing a model.
By Soroosh Tayebi Arasteh, Sebastian Ziegelmayer, Mahshad Lotfinia, Lisa Adams, Sven Nebelung, Jakob Nikolas Kather, Daniel Truhn
The paper introduces SCCL-DT, a method that improves zero‑shot chest X‑ray diagnosis by addressing semantic conflicts in contrastive vision‑language pre‑training. It uses adaptive thresholds derived from exponential‑moving‑average statistics to hard‑mask conflicting pairs, and combines two teachers—an inclusive and a consensus rule—through distillation into a single student model. On five public chest X‑ray benchmarks, SCCL‑DT achieves a mean zero‑shot AUC of 0.850, outperforming the best baseline at 0.837.
By Ting Liu, Haoran Lai, HaobinChen, Ming Liu, S Kevin Zhou
We describe the DS@GT submissions to the ImageCLEFmedical Caption 2026 challenge, which continues a long-running benchmark on the ROCOv2 dataset with two tracks: Concept Detection (Task 1), assigning UMLS Concept Unique Identifiers (CUIs) to radiology images, and Caption Prediction (Task 2), generating natural-language captions. For Task 1, our primary submission was a three-way late-fusion ensemble of ConvNeXt-V2, BiomedCLIP ViT-B/16, and DenseNet-169 with a regularized ''Honest Threshold Tuning'' procedure designed to avoid validation overfitting on rare concepts; this submission ranked first on the official submission with a primary $F_1$ of $0.
The paper introduces Similarity-as-Evidence (SaE), a framework that calibrates overconfident Vision‑Language Models (VLMs) for medical active learning by reinterpreting text‑image similarity vectors as evidence and modeling them with a Dirichlet distribution. SaE’s Similarity Evidence Head (SEH) captures uncertainty through vacuity and dissonance, enabling a dual‑factor acquisition strategy that prioritizes high‑vacuity samples early and high‑dissonance samples later. Experiments on ten public medical imaging datasets demonstrate state‑of‑the‑art macro‑averaged accuracy (82.57%) and improved calibration (NLL = 0.425) under a 20% label budget.
By Zhuofan Xie, Zishan Lin, Jinliang Lin, Jie Qi, Shaohua Hong, Shuo Li
arXiv:2607. 27763v1 Announce Type: cross Abstract: We describe the DS@GT submissions to the ImageCLEFmedical Caption 2026 challenge, which continues a long-running benchmark on the ROCOv2 dataset with two tracks: Concept Detection (Task 1), assigning UMLS Concept Unique Identifiers (CUIs) to radiology images, and Caption Prediction (Task 2), generating natural-language captions.
By Bowen Wang, Youwen Zhang, Ritesh Mehta