IEEE Journal of Biomedical and Health Informatics最新文献

{"title":"Meta-Entity Driven Triplet Mining for Aligning Medical Vision-Language Models","authors":"Melih B. Yilmaz;Saban Ozturk;Muti Kara;M. Talat Yavuz;Ekim Gumeler;Aykut Koç;Tolga Çukur","doi":"10.1109/JBHI.2026.3678306","DOIUrl":"10.1109/JBHI.2026.3678306","url":null,"abstract":"Imaging-based diagnostics rely on evaluation of both medical images and radiology reports, but increasing data volumes strain medical experts, leading to errors and workflow delays. Medical vision-language models (med-VLMs) offer an efficient approach for processing multimodal imaging data, especially for chest X-rays (CXRs), though their success depends on effective image-text alignment. Existing alignment methods for med-VLMs, primarily based on contrastive learning, often focus on coarse-grained disease class separation and overlook fine-grained pathology attributes such as location, size, or severity, which results in suboptimal representations. We introduce MedTrim (<underline>M</u>eta-<underline>e</u>ntity-<underline>d</u>riven <underline>Tri</u>plet <underline>m</u>ining), a novel alignment method that improves precision via structured triplet learning guided by meta-entities extracted from radiology reports. Unlike conventional contrastive and triplet frameworks for representational learning that rely on global class labels or implicit similarity references, MedTrim explicitly models hierarchical relationships between pathology attributes to preserve clinically meaningful intra-class variation. To do this, MedTrim leverages a domain-specific ontology to identify adjectival qualifiers and directional descriptors of pathology, a novel entity-aware triplet mining score to capture hierarchical inter-sample similarity, and a multimodal alignment objective that enforces consistency across image-text pairs sharing detailed pathology attributes without compromising within-modality relationships. MedTrim improves performance in downstream retrieval, classification, and generation tasks compared to leading alignment methods.","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"30 10","pages":"8799-8812"},"PeriodicalIF":7.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11457238","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147528444","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"HDPL: Hypergraph-Based Dynamic Prompting Learning for Incomplete Multimodal Medical Learning","authors":"Xiaomin Zhou;Guoheng Huang;Qin Zhao;Jianbin He;Xiaochen Yuan;Ming Li;Chi-Man Pun;Ling Guo;Baiying Lei;Qi Yang","doi":"10.1109/JBHI.2026.3671347","DOIUrl":"10.1109/JBHI.2026.3671347","url":null,"abstract":"Multimodal learning has garnered significant attention in the medical field due to its ability to provide a more comprehensive perspective utilizing various types of data, that aids in making more accurate decisions. However, the complexity of medical data, coupled with missing modalities, severely hinders predictive accuracy. Existing methods for multimodal learning with missing modalities still face considerable challenges. For instance, approaches that construct multimodal shared feature spaces often result in high computational costs, while methods that infer missing modalities based on complete ones may overly rely on the complete modalities, potentially skewing results. Pre-trained transformer methods address these issues but still have limitations, such as it can only process one missing modality at testing-stage. This is partly because structured data, unlike sequential data, lacks inherent minimum semantic units or natural order. Additionally, the positional encodings generated by this type of methods may introduce information interference when applied to structured data, leading to poor alignment with sequential data during modality fusion in transformer models. To tackle these challenges, we introduce HDPL: Hypergraph-based Dynamic Prompt Learning for Incomplete Multimodal Medical Learning, comprising three modules. The High-Order Hypergraph Embedding module can identify the minimal semantic units within structured data and utilizes hypergraph structures to extract high-dimensional features from clinical data. The Multimodal Medical Data Integrator module closes the distance of the embedding vectors corresponding in the shared space of modality-features, facilitating the integration of modalities in transformer. The Dynamic Network Structure Optimization module is a dynamic learning network by dynamically change the width and depth of network, improving the overall performance of the model, and it alleviates the shortcomings caused by incomplete modality to some extent. Through comprehensive experimentation, we demonstrate the efficiency and robustness of our model in dealing missing modalities and reducing training-burdens.","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"30 10","pages":"8417-8430"},"PeriodicalIF":7.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147389891","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Unsupervised Hybrid Framework for ANomaly Detection (HAND) - Applied to Screening Mammogram","authors":"Zhemin Zhang;Bhavika Patel;Bhavik Patel;Imon Banerjee","doi":"10.1109/JBHI.2025.3638713","DOIUrl":"10.1109/JBHI.2025.3638713","url":null,"abstract":"Out-of-distribution (OOD) detection is essential for improving the generalization of of AI models used in mammogram screening, as unknown distribution shifts in external datasets can degrade performance. Identifying OOD samples helps maintain strong model performance across internal-external datasets. Given the challenge of limited prior knowledge about OOD samples in external datasets, unsupervised generative learning is a preferable solution which trains the model to discern the normal characteristics of in-distribution (ID) data. Generic hypothesis is that during inference, the model aims to reconstruct ID samples accurately, while OOD samples exhibit poorer reconstruction due to their divergence from normality. Inspired by the hybrid architectures combining CNNs and transformers, we developed a novel backbone - HAND, for detecting OOD from large-scale digital screening mammogram studies. However, relying solely on reconstruction error is insufficient for reliable OOD detection, as it may not consistently distinguish between ID and OOD samples. To address this, we incorporated synthetic OOD samples and a parallel discriminator in the latent space to distinguish between ID and OOD samples. Additionally, we applied gradient reversal to the OOD reconstruction loss, effectively discouraging the model from accurately reconstructing OOD inputs and reinforcing its ability to differentiate them from ID samples. An anomaly score is computed by weighting the reconstruction and discriminator loss. On internal RSNA mammogram held-out test and external Mayo clinic hand-curated dataset, the proposed HAND model outperformed encoder-based and GAN-based baselines, and interestingly, it also outperformed the hybrid CNN+transformer baselines. Therefore, the proposed HAND pipeline offers an automated efficient computational solution for domain-specific quality checks in external screening mammograms, yielding actionable insights without direct exposure to the private medical imaging data.","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"30 10","pages":"8861-8869"},"PeriodicalIF":7.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145632887","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GL-Fusion: A Multi-Omics Integration Method Based on Graph-Level Structure Fusion and Locus-Level Feature Fusion for Cancer Subtype Classification","authors":"Kaiwen Tan;Qi Wei;Min Luo;Honghao Zhu;Chun Jiang;Zhenqiu Shu;Jianqiu Kong","doi":"10.1109/JBHI.2025.3624554","DOIUrl":"10.1109/JBHI.2025.3624554","url":null,"abstract":"The rapid development of multi-omics data has provided new opportunities for cancer subtype classification. Due to the ability to model gene associations by constructing graph among genes, multi-omics cancer classification based on graphs has attracted considerable attention from researchers. However, due to the high-dimensional nature of multi-omics data, existing graph construction methods based on cosine similarity may lead to noise and low-quality edges. Moreover, considering the complexity of associations within and across omics, existing methods either focus solely on multi-omics fusion at the graph structure level or only consider multi-omics fusion at the representation level, making it challenging to comprehensively model this complex relationship. To this end, we propose a novel multi-omics integration method that combines graph-level structure fusion and locus-level feature fusion to enhance the performance of cancer subtype classification (GL-Fusion). In the graph-level structure fusion module, we integrate multi-omics gene-gene graphs using similarity network fusion method and optimize the graph structure with structural entropy and the protein-protein interaction network. In the locus-level feature fusion module, we employ a locus-level graph convolutional network to integrate multi-omics features, learn gene-level fused representations, and predict cancer subtypes. Empirical validation across four publicly accessible datasets (BRCA, HNSC, LGG, THCA) indicated that our method achieved superior performance compared to 12 representative multi-omics cancer subtyping methods. Ablation experiments, experiments with different omics combinations, and evaluation experiments on modeling associations within and across omics were used to further validate the effectiveness of the method.","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"30 10","pages":"8204-8217"},"PeriodicalIF":7.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145722684","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DGRTA: Cross-Modality Unsupervised Domain Adaptation for Intracranial Vessel Segmentation via Dual-Gated Refinement and Topology-Aware Weighting","authors":"Yijia Zheng;Jiahui Lv;Yaping Wu;Xinsheng Mao;Chao Zheng;Meiyun Wang;Hua Guo","doi":"10.1109/JBHI.2026.3670538","DOIUrl":"10.1109/JBHI.2026.3670538","url":null,"abstract":"TOF-MRA intracranial vessel segmentation is critical in clinical practice but challenged by limited annotations and significant cross-modality domain shifts. To address these issues, this study proposes DGRTA, an unsupervised domain adaptation framework that integrates cross-modality pseudo-label generation, a dual-gated pseudo-label refinement strategy (DGR), and a topology-aware weighting mechanism (TA). Initially, rigid and non-rigid registration are used to transfer CTA predictions to TOF-MRA to generate initial pseudo-labels. DGR then refines these labels using prediction probabilities and image intensity, enhancing sensitivity and specificity, while TA leverages Persistence Diagrams (PD) to quantify topological discrepancies and dynamically adjust loss weights. Experiments on 185 paired CTA/TOF-MRA cases demonstrated that DGRTA consistently improved performance across four backbone architectures (UNet, Attention UNet, UNETR, Swin UNETR). The Attention UNet-DGRTA achieved the best results, with a Dice of 0.810, clDice of 0.800, and an AHD of 0.353 mm on the validation set, significantly outperforming the baseline model (p < 0.001). DGRTA offers a feasible solution that reduces reliance on extensive manual annotations, underscoring the potential of unsupervised cross-modality segmentation in various vascular imaging applications.","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"30 10","pages":"8850-8860"},"PeriodicalIF":7.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147389923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"IEEE Journal of Biomedical and Health Informatics Information for Authors","authors":"","doi":"10.1109/JBHI.2026.3733799","DOIUrl":"https://doi.org/10.1109/JBHI.2026.3733799","url":null,"abstract":"","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"30 10","pages":"C3-C3"},"PeriodicalIF":7.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11720473","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097914","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Smart Foot Wearables: An Interdisciplinary Perspective.","authors":"Kamen Ivanov, Md Babul Islam, Dimitar Dimov, Tanya Malakova, Todor Stoyanov, Zhanyong Mei, Ludwig Lubich, Diana Kichukova, Nguyen Van Sau, Aneliya Kostadinova, Teodor Milenov, Nikolay Donkov, Yushi Liu, Yan Yan, Lei Wang, Antonio Guerrieri, Raffaele Gravina, Qimeng Li, Giancarlo Fortino","doi":"10.1109/JBHI.2026.3737611","DOIUrl":"https://doi.org/10.1109/JBHI.2026.3737611","url":null,"abstract":"<p><p>Smart foot wearables (SFWs) are emerging components of Internet of Medical Things (IoMT) ecosystems and wearable cyber-physical platforms. They support mobility assessment, rehabilitation monitoring, foot-health management, fall-risk assessment, sports biomechanics, navigation, and biometric authentication. Unlike stationary instrumentation, SFWs face coupled constraints in power, form factor, near-ground wireless connectivity, and long-term mechanical and thermal loading. This survey presents a cross-layer synthesis of SFW systems. The proposed taxonomy defines five primary classification axes: form factor and fixation regime; sensing stack and plantar semantics; sampling, timing, and bilateral alignment; computation and communication partition; and validation and transfer setting. It further identifies three cross-cutting deployment modifiers: sensor-stack mechanics and drift sources; energy, charging, and maintenance regime; and feedback and closed-loop safety. The survey analyzes sensing modalities, communication constraints, and data-processing pipelines, with particular attention to drift, timing integrity, multimodal fusion, and transfer across footwear and terrain. It also examines clinical-data interoperability, privacy and security, charging and calibration burden, washability, durability, and other barriers to sustained deployment. Literature was identified through Google Scholar, six bibliographic databases, citation tracking, and targeted supplementation. The search channels, search period, shared concept blocks, and ScienceDirect adaptation are reported in the methodology. Across the heterogeneous evidence reviewed, the synthesis suggests that real-world validity depends less on isolated sensor accuracy than on interactions among calibration stability, timing integrity, wireless delivery, energy autonomy, validation design, and maintenance. The survey concludes with reporting and research priorities for robust, clinically relevant, and maintainable SFW systems.</p>","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"PP ","pages":""},"PeriodicalIF":7.7,"publicationDate":"2026-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149108874","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Robust Non-Invasive Monitoring of Cumulative Tibia Load in Male Recreational Runners Using Multi-Modal Wearables and Deep Learning.","authors":"Zifan Xia, Yihan Qian, Dong Sun, Chengyuan Zhu, Yufan Xu, Xuanzhen Cen, Yang Song, Zixiang Gao, Liangliang Xiang, Monem Jemni, Yaodong Gu","doi":"10.1109/JBHI.2026.3736502","DOIUrl":"https://doi.org/10.1109/JBHI.2026.3736502","url":null,"abstract":"<p><p>Tibial stress fractures are overuse injuries in runners and are associated with cumulative internal bone stress. Current wearable devices usually measure external biomechanical loads rather than tissue-level stress within the tibia. We developed and evaluated a wearable-sensor-based surrogate model for reconstructing the temporal evolution of tibial bone stress in healthy male recreational runners, using participant-specific dynamic finite element analysis (FEA) as the supervision source. Dynamic FEA-derived stress fields from 50 male runners, with heterogeneous and orthotropic material properties, were used as ground-truth regression targets for deep learning. The main model used synchronized inertial measurement unit and surface electromyography signals as inputs and a convolutional-transformer backbone to map surface signals to dynamic tibial stress outputs. A self-correction with adaptive mask training module was also evaluated as an auxiliary regularization strategy. The model predicted temporal stress waveforms (R² = 0.90) and global peak magnitudes (Pearson r = 0.944), with low systematic bias but residual individual-level variability. Robustness training was evaluated under sensor noise and modality-loss conditions, and edge-device inference required less than 60 ms. These findings indicate that wearable sensors can support tibial stress reconstruction in male recreational runners, although external validation in female runners and other populations is required before broader application.</p>","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"PP ","pages":""},"PeriodicalIF":7.7,"publicationDate":"2026-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149136025","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"L2DiffM: LLM-Guided Variational Expert Diffusion for Target-Aware 3D Molecular Generation.","authors":"Jiahao Liu, Yongxuan Lai, Jie Xia, Fan Yang","doi":"10.1109/JBHI.2026.3737835","DOIUrl":"https://doi.org/10.1109/JBHI.2026.3737835","url":null,"abstract":"<p><p>Target-aware three-dimensional (3D) molecular generation aims to produce candidate molecules with chemical validity, conformational plausibility, and target compatibility under protein binding pocket constraints. Existing pocket conditioned diffusion models rely primarily on local geometric and atom-level state modeling, leaving high-level medicinal chemistry semantics underutilized; moreover, single-path de noising and fixed-step sampling strategies fail to accommo date the heterogeneity of diffusion states and inter-sample structural diversity, limiting generation quality and structural convergence stability. To address these limitations, we propose L2DiffM, a large language model (LLM)-guided variational expert diffusion framework for target-aware 3D molecular generation. Specifically, L2DiffM establishes a semantic transfer pathway from a language teacher to a protein pocket conditioned structural prior by learning high-level medicinal chemistry patterns in 1D molecular representations, providing pharmacochemical constraints during late-stage denoising. Expert selection is formulated as latent variable inference, yielding a variational self-routed mixture-of-experts (VSRD MoE) denoising mechanism that simultaneously adapts to diffusion-state heterogeneity and inter-sample structural variability via energy- and uncertainty-guided responsibility al location, mitigating the averaging bias of shared denoisers. We further propose Bond-Critical Adaptive Sampling (BCAS), which jointly evaluates bond-topology uncertainty, local geo metric uncertainty, and pocket interaction sensitivity to adap tively reallocate the sampling budget and apply sparse local corrections at high-risk boundary states. Experiments on the PDBbind and CrossDocked benchmarks demonstrate consistent improvements in molecular validity, structural stability, binding affinity, and conformational quality, confirming that L2DiffM provides an effective modeling paradigm for target aware 3D molecular generation.</p>","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"PP ","pages":""},"PeriodicalIF":7.7,"publicationDate":"2026-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149108811","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"InFo-DRP: Integrating Invariant Learning and Foundation Representations for Unseen Drug Response Prediction.","authors":"Jisang Lee, Taelim Yeom, June-Woo Kim, Hyunju Lee","doi":"10.1109/JBHI.2026.3738922","DOIUrl":"https://doi.org/10.1109/JBHI.2026.3738922","url":null,"abstract":"<p><p>Accurate prediction of cancer cell responses to chemical compounds remains a fundamental challenge in precision oncology and drug discovery. Although several drug response prediction (DRP) models have been proposed, most were evaluated using random splits of drug-cell line pairs, allowing the same drugs to appear in both training and test sets. In practice, models must predict responses to previously unseen compounds, necessitating robustness under drug- blind and scaffold- blind splits. However, many existing models show substantial performance degradation under these conditions, indicating limited robustness to chemical distribution shift. To address this challenge, we propose InFo-DRP that integrates invariant learning with foundation-model drug representations. InFo-DRP incorporates multi-branch drug encoders (MolBERT, Chemformer, and ECFP4) with physicochemical descriptors, separates drug and cell representations into invariant and spurious components, and applies dual-path fusion for prediction. Across two Genomics of Drug Sensitivity in Cancer (GDSC) benchmarks, InFo-DRP outperforms strong baselines under drug- blind and scaffold- blind evaluations. On five non-cancerous cell lines, InFo-DRP maintains predictive performance under scaffold shift, suggesting potential utility beyond cancer-specific settings. In external validation on the Cancer Cell Line Encyclopedia dataset, InFo-DRP achieves the highest correlation among the compared methods without any adaptation. Overall, InFo-DRP provides a practical approach for generalizing DRP to previously unseen chemical space.</p>","PeriodicalId":13073,"journal":{"name":"IEEE Journal of Biomedical and Health Informatics","volume":"PP ","pages":""},"PeriodicalIF":7.7,"publicationDate":"2026-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149136017","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}