Journal of Biomedical Informatics最新文献_第2页

RINet: synthetic data training for indirect estimation of clinical reference distributions RINet：用于间接估计临床参考分布的综合数据训练。

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2026-01-08 DOI: 10.1016/j.jbi.2026.104980

Jack LeBien , Julian Velev , Abiel Roche-Lima

Background

Indirect methods for estimating clinical reference intervals (RIs) use statistical analysis to identify non-pathological sub-distributions within large datasets acquired from routine clinical testing. This approach has the potential to accelerate the estimation of precise RIs, accounting for influential variables such as age, gender, and ethnicity. Most existing methods are based on traditional statistics and hand-crafted algorithms. The investigation of supervised learning, which often outperforms traditional approaches, has been impeded by the limitations of real-world data. However, previous studies have widely used synthetic data for evaluating and benchmarking indirect methods due several advantages over real-world data, including greater control, variability, accessibility, and the availability of exact ground-truth RIs. Synthetic data may also provide a pathway for developing data-driven solutions for indirect RI estimation.

Methods

In this study, we leveraged synthetic data to train two convolutional neural networks (CNNs) to predict the parameters of underlying reference distributions (RDs) in diverse real-world clinical datasets. While one model was trained for standard univariate data, the other was extended to bivariate data, enabling the prediction of covariance between clinical analytes. Trained models were evaluated using both real-world and synthetic test datasets and compared with four alternative algorithms.

Results

Model predictions closely matched directly estimated RIs and RDs in real-world data and known RDs in synthetic data, outperforming four alternative indirect methods: GMM, refineR, reflimR, and RINet_v1. Using labeled healthy and HCV-positive groups in real data, we compared established univariate RIs with predicted multivariate reference regions (MRRs). On average, the MRRs showed 1) higher coverage of healthy patients (closer to the desired 95%) and 2) smaller regions, which reduce the likelihood of including abnormal values.

Conclusions

Synthetic data training is a viable approach for developing accurate indirect RI estimation models for both univariate and bivariate clinical data. This strategy could help address some limitations of real-world data, direct analyses, and univariate RIs.

背景：估计临床参考区间（RIs）的间接方法使用统计分析来识别从常规临床检测获得的大型数据集中的非病理亚分布。考虑到年龄、性别和种族等有影响的变量，这种方法有可能加速对精确RIs的估计。大多数现有的方法都是基于传统的统计和手工制作的算法。监督学习的研究通常优于传统方法，但受到现实世界数据的限制。然而，先前的研究已经广泛使用合成数据来评估和对间接方法进行基准测试，因为与真实世界的数据相比，合成数据具有一些优势，包括更大的可控性、可变性、可访问性和精确的真实RIs的可用性。合成数据还可以为开发数据驱动的间接RI估计解决方案提供途径。方法：在这项研究中，我们利用合成数据来训练两个卷积神经网络（cnn）来预测不同现实世界临床数据集中潜在参考分布（rd）的参数。当一个模型被训练为标准的单变量数据时，另一个模型被扩展到双变量数据，从而能够预测临床分析者之间的协方差。训练后的模型使用真实世界和合成测试数据集进行评估，并与四种替代算法进行比较。结果：模型预测与实际数据中直接估计的RIs和rd以及合成数据中的已知rd密切匹配，优于四种替代间接方法：GMM， refineR， reflimR和RINetv1。使用真实数据中标记的健康组和hcv阳性组，我们比较了已建立的单变量RIs与预测的多变量参考区域（MRRs）。平均而言，磁共振成像显示1)健康患者的覆盖率更高（接近预期的95%），2)区域更小，这降低了包括异常值的可能性。结论：综合数据训练是为单变量和双变量临床数据建立准确的间接RI估计模型的可行方法。这种策略可以帮助解决现实世界数据、直接分析和单变量RIs的一些限制。

{"title":"RINet: synthetic data training for indirect estimation of clinical reference distributions","authors":"Jack LeBien , Julian Velev , Abiel Roche-Lima","doi":"10.1016/j.jbi.2026.104980","DOIUrl":"10.1016/j.jbi.2026.104980","url":null,"abstract":"<div><h3>Background</h3><div>Indirect methods for estimating clinical reference intervals (RIs) use statistical analysis to identify non-pathological sub-distributions within large datasets acquired from routine clinical testing. This approach has the potential to accelerate the estimation of precise RIs, accounting for influential variables such as age, gender, and ethnicity. Most existing methods are based on traditional statistics and hand-crafted algorithms. The investigation of supervised learning, which often outperforms traditional approaches, has been impeded by the limitations of real-world data. However, previous studies have widely used synthetic data for evaluating and benchmarking indirect methods due several advantages over real-world data, including greater control, variability, accessibility, and the availability of exact ground-truth RIs. Synthetic data may also provide a pathway for developing data-driven solutions for indirect RI estimation.</div></div><div><h3>Methods</h3><div>In this study, we leveraged synthetic data to train two convolutional neural networks (CNNs) to predict the parameters of underlying reference distributions (RDs) in diverse real-world clinical datasets. While one model was trained for standard univariate data, the other was extended to bivariate data, enabling the prediction of covariance between clinical analytes. Trained models were evaluated using both real-world and synthetic test datasets and compared with four alternative algorithms.</div></div><div><h3>Results</h3><div>Model predictions closely matched directly estimated RIs and RDs in real-world data and known RDs in synthetic data, outperforming four alternative indirect methods: GMM, <em>refineR</em>, <em>reflimR</em>, and RINet<sub>v1</sub>. Using labeled healthy and HCV-positive groups in real data, we compared established univariate RIs with predicted multivariate reference regions (MRRs). On average, the MRRs showed 1) higher coverage of healthy patients (closer to the desired 95%) and 2) smaller regions, which reduce the likelihood of including abnormal values.</div></div><div><h3>Conclusions</h3><div>Synthetic data training is a viable approach for developing accurate indirect RI estimation models for both univariate and bivariate clinical data. This strategy could help address some limitations of real-world data, direct analyses, and univariate RIs.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"174 ","pages":"Article 104980"},"PeriodicalIF":4.5,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145948483","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}

引用次数: 0

A computational framework for predicting drug-target interactions by fusing gene ontology information with cross attention 交叉关注融合基因本体信息预测药物-靶标相互作用的计算框架

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2026-01-02 DOI: 10.1016/j.jbi.2025.104976

Wenchao Cui, Pingjian Ding, Lingyun Luo, Shunheng Zhou, Hui Jiang

Motivation

Identifying drug–target interactions (DTIs) is a critical step in both drug discovery and drug repurposing. Accurate in silico prediction of DTIs can substantially reduce development time and costs. Recent advances in sequence-based methods have leveraged attention mechanisms to improve prediction accuracy. However, these approaches typically rely solely on the molecular structures of drugs and proteins, overlooking higher-level semantic information that reflects functional and biological relationships.

Results

In this work, we propose GODTI, a novel Gene Ontology-guided Drug-Target Interaction prediction model that enhances the performance through multimodal feature integration. GODTI comprises three major components: a feature extraction module, a multimodal fusion module, and an intermolecular interaction modeling module. In the protein feature extractor, both functional descriptors derived from Gene Ontology and sequence-based embeddings from amino acid sequences are obtained and combined. These protein representations are then integrated with drug molecular features via the multimodal fusion module and subsequently processed by the interaction modeling module to predict potential interactions. We evaluated GODTI under four realistic experimental settings, demonstrating consistent improvements over state-of-the-art baselines. Furthermore, case studies validated the practical utility of GODTI in accurately identifying novel, low-cost DTIs, underscoring its potential to accelerate drug discovery workflows.

动机识别药物-靶标相互作用（DTIs）是药物发现和药物再利用的关键步骤。准确的dti计算机预测可以大大减少开发时间和成本。基于序列的方法的最新进展利用注意机制来提高预测的准确性。然而，这些方法通常只依赖于药物和蛋白质的分子结构，而忽略了反映功能和生物关系的更高层次的语义信息。结果提出了一种基于基因本体论的药物-靶标相互作用预测模型GODTI，该模型通过多模态特征集成提高了药物-靶标相互作用预测的性能。GODTI包括三个主要部分：特征提取模块、多模态融合模块和分子间相互作用建模模块。在蛋白质特征提取器中，获得了来自基因本体的功能描述子和来自氨基酸序列的基于序列的嵌入子并进行了组合。然后通过多模态融合模块将这些蛋白质表征与药物分子特征整合，随后由相互作用建模模块进行处理，以预测潜在的相互作用。我们在四种现实的实验设置下评估了GODTI，显示出与最先进的基线相一致的改进。此外，案例研究证实了GODTI在准确识别新型低成本dti方面的实际效用，强调了其加速药物发现工作流程的潜力。

{"title":"A computational framework for predicting drug-target interactions by fusing gene ontology information with cross attention","authors":"Wenchao Cui, Pingjian Ding, Lingyun Luo, Shunheng Zhou, Hui Jiang","doi":"10.1016/j.jbi.2025.104976","DOIUrl":"10.1016/j.jbi.2025.104976","url":null,"abstract":"<div><h3>Motivation</h3><div>Identifying drug–target interactions (DTIs) is a critical step in both drug discovery and drug repurposing. Accurate <em>in silico</em> prediction of DTIs can substantially reduce development time and costs. Recent advances in sequence-based methods have leveraged attention mechanisms to improve prediction accuracy. However, these approaches typically rely solely on the molecular structures of drugs and proteins, overlooking higher-level semantic information that reflects functional and biological relationships.</div></div><div><h3>Results</h3><div>In this work, we propose GODTI, a novel Gene Ontology-guided Drug-Target Interaction prediction model that enhances the performance through multimodal feature integration. GODTI comprises three major components: a feature extraction module, a multimodal fusion module, and an intermolecular interaction modeling module. In the protein feature extractor, both functional descriptors derived from Gene Ontology and sequence-based embeddings from amino acid sequences are obtained and combined. These protein representations are then integrated with drug molecular features via the multimodal fusion module and subsequently processed by the interaction modeling module to predict potential interactions. We evaluated GODTI under four realistic experimental settings, demonstrating consistent improvements over state-of-the-art baselines. Furthermore, case studies validated the practical utility of GODTI in accurately identifying novel, low-cost DTIs, underscoring its potential to accelerate drug discovery workflows.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"174 ","pages":"Article 104976"},"PeriodicalIF":4.5,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145891182","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}

引用次数: 0

Fusion framework: Conditional-aware one-stage nested event extraction model 融合框架：条件感知的单阶段嵌套事件提取模型。

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2026-01-01 DOI: 10.1016/j.jbi.2025.104972

Sen Niu , Xiaohong Han , Liu Cao , Ye Tian , Ding Yuan , Longlong Cheng

We present CA-NEE, a Conditional-Aware one-stage model for overlapping and nested biomedical event extraction. CA-NEE integrates an event-type-aware conditioning mechanism with token-pair relation modeling to jointly identify triggers, argument spans, and roles. A Conditional Layer Normalization (CLN) dynamically adapts token representations to candidate event types, and a parallel word-pair scorer predicts span boundaries and roles in a single pass. Evaluations on GENIA11 and GENIA13 show consistent gains in Trigger Classification (TC) and Argument Classification (AC) over strong baselines, particularly on complex overlapping and nested structures. These results demonstrate that CA-NEE offers an effective and efficient solution for biomedical event extraction.

我们提出了一种用于重叠和嵌套生物医学事件提取的条件感知单阶段模型CA-NEE。CA-NEE将事件类型感知的条件调节机制与令牌对关系建模集成在一起，以联合识别触发器、参数范围和角色。条件层规范化（Conditional Layer Normalization， CLN）动态地将令牌表示适应候选事件类型，并行词对评分器在一次传递中预测跨度边界和角色。对GENIA11和GENIA13的评估显示，在强基线上，触发分类（TC）和参数分类（AC）取得了一致的进展，特别是在复杂的重叠和嵌套结构上。这些结果表明CA-NEE为生物医学事件提取提供了有效的解决方案。

引用次数: 0

Reviewer Acknowledgement 2025 审稿人致谢2025。

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2026-01-01 DOI: 10.1016/j.jbi.2025.104974

引用次数: 0

Integrating retrospective quality assessment with real-time guideline application to support the episodic application of clinical guidelines over significant time periods 将回顾性质量评估与实时指南应用相结合，支持临床指南在重要时间段内的阶段性应用。

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2026-01-01 DOI: 10.1016/j.jbi.2025.104975

Bruria Ben Shahar , Yuval Shahar , Shai Jaffe , Odeya Cohen , Erez Shalom , Maya Selivanova , Ephraim Rimon , Irit Hochberg , Ayelet Goldstein

Background

Evidence-based clinical guidelines (GLs) are essential for standardizing care, yet often difficult to apply. Most clinical decision support systems (CDSSs) assume continuous application, which misaligns with the episodic nature of real-world workflows.

Objectives

To design, implement, and evaluate e-Picard, a CDSS that provides GL-based recommendations through episodic, intermittent, on-demand consultations. The system supports retrospective assessment of past care and prospective identification of required actions. The evaluation focused on system validity and on its potential, in a retrospective simulation on real-world data, to enhance staff adherence to the GLs and to assess the potential effect of varying the frequency of the consultations.

Methods

The system development involved three preprocessing steps: (1) acquisition of free-text GLs with domain experts; (2) modeling procedural logic as workflows; and (3) flattening these into declarative temporal patterns for retrospective quality assessment and prospective recommendations. At runtime, e-Picard analyzes offline patient data to identify missed actions, computes compliance using fuzzy logic, and generates context-specific recommendations. e-Picard was applied to pressure-ulcer (PU) and diabetes management (DM) GLs, adapted for episodic use. Technical validation was performed on records from 43 PU and 82 DM patients. A retrospective simulation using 1,000 patients per domain estimated potential increases in adherence under varying consultation frequencies.

Results

Technical manual validation showed high correctness (≥99 %) and completeness (up to 98 %), based on 3,110 PU and 12,538 DM data instances (i.e., clinical measurements or actions), across various clinical scenarios over two-week observation periods.

Retrospective simulation covered 57,860 PU and 100,940 DM data instances with estimated adherence potentially increasing from 68 %–69 % to 89 %–97 % for PU and from 14 %–15 % to 60 %–87 % for DM, in the real-world data retrospective simulation, assuming full adherence of the staff to the system’s recommendations, depending on the scenario. Higher consultation frequency yielded greater gains, and adherence variability across hospital units and patient subgroups was reduced.

Conclusions

Episodic CDSSs can deliver accurate, context-aware recommendations in environments with intermittent use and incomplete data, with the potential, assuming that the real-world data retrospective simulation results hold, to enhance adherence and consistency in care.

背景：循证临床指南（GLs）是标准化护理必不可少的，但往往难以应用。大多数临床决策支持系统（cdss）假设持续应用，这与现实世界工作流程的偶然性不一致。目的：设计、实施和评估e-Picard，这是一个CDSS，通过偶发的、间歇的、按需咨询提供基于gl的建议。该系统支持对过去护理的回顾性评估和对所需行动的前瞻性识别。评价的重点是系统有效性及其潜力，通过对真实世界数据的回顾性模拟，加强工作人员对全球目标的遵守，并评估改变咨询频率的潜在影响。方法：系统开发包括三个预处理步骤：(1)与领域专家一起获取自由文本GLs；(2)将过程逻辑建模为工作流；(3)将这些扁平化为陈述性时间模式，用于回顾性质量评估和前瞻性建议。在运行时，e-Picard分析离线患者数据以识别遗漏的操作，使用模糊逻辑计算遵从性，并生成特定于上下文的建议。e-Picard应用于压疮（PU）和糖尿病管理（DM） GLs，适用于间歇性使用。对43例PU和82例DM患者的记录进行了技术验证。每个领域使用1000名患者的回顾性模拟估计了不同咨询频率下依从性的潜在增加。结果：技术手册验证显示高正确性（≥99 %）和完整性（高达98 %），基于3,110 PU和12,538 DM数据实例（即临床测量或行动），跨越两周观察期的各种临床场景。在现实世界的数据回顾性模拟中，假设员工完全遵守系统的建议，根据具体情况，回顾性模拟涵盖了57,860个PU和100,940个DM数据实例，估计依从性可能从PU的68 %-69 %增加到89 %-97 %，DM的14 %-15 %增加到60 %-87 %。更高的咨询频率产生更大的收益，并且降低了医院单位和患者亚组之间的依从性差异。结论：偶发性cdss可以在间歇性使用和数据不完整的环境中提供准确的、情境感知的建议，假设真实世界数据回顾性模拟结果有效，具有增强护理依从性和一致性的潜力。

{"title":"Integrating retrospective quality assessment with real-time guideline application to support the episodic application of clinical guidelines over significant time periods","authors":"Bruria Ben Shahar , Yuval Shahar , Shai Jaffe , Odeya Cohen , Erez Shalom , Maya Selivanova , Ephraim Rimon , Irit Hochberg , Ayelet Goldstein","doi":"10.1016/j.jbi.2025.104975","DOIUrl":"10.1016/j.jbi.2025.104975","url":null,"abstract":"<div><h3>Background</h3><div>Evidence-based clinical guidelines (GLs) are essential for standardizing care, yet often difficult to apply. Most clinical decision support systems (CDSSs) assume continuous application, which misaligns with the episodic nature of real-world workflows.</div></div><div><h3>Objectives</h3><div>To design, implement, and evaluate e-Picard, a CDSS that provides GL-based recommendations through episodic, intermittent, on-demand consultations. The system supports retrospective assessment of past care and prospective identification of required actions. The evaluation focused on system validity and on its potential, in a retrospective simulation on real-world data, to enhance staff adherence to the GLs and to assess the potential effect of varying the frequency of the consultations.</div></div><div><h3>Methods</h3><div>The system development involved three preprocessing steps: (1) acquisition of free-text GLs with domain experts; (2) modeling procedural logic as workflows; and (3) flattening these into declarative temporal patterns for retrospective quality assessment and prospective recommendations. At runtime, e-Picard analyzes offline patient data to identify missed actions, computes compliance using fuzzy logic, and generates context-specific recommendations. e-Picard was applied to pressure-ulcer (PU) and diabetes management (DM) GLs, adapted for episodic use. Technical validation was performed on records from 43 PU and 82 DM patients. A retrospective simulation using 1,000 patients per domain estimated potential increases in adherence under varying consultation frequencies.</div></div><div><h3>Results</h3><div>Technical manual validation showed high correctness (≥99 %) and completeness (up to 98 %), based on 3,110 PU and 12,538 DM data instances (i.e., clinical measurements or actions), across various clinical scenarios over two-week observation periods.</div><div>Retrospective simulation covered 57,860 PU and 100,940 DM data instances with estimated adherence potentially increasing from 68 %–69 % to 89 %–97 % for PU and from 14 %–15 % to 60 %–87 % for DM, in the real-world data retrospective simulation, assuming full adherence of the staff to the system’s recommendations, depending on the scenario. Higher consultation frequency yielded greater gains, and adherence variability across hospital units and patient subgroups was reduced.</div></div><div><h3>Conclusions</h3><div>Episodic CDSSs can deliver accurate, context-aware recommendations in environments with intermittent use and incomplete data, with the potential, assuming that the real-world data retrospective simulation results hold, to enhance adherence and consistency in care.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"173 ","pages":"Article 104975"},"PeriodicalIF":4.5,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145878473","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}

引用次数: 0

Journal's cover 杂志的封面

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2026-01-01 DOI: 10.1016/S1532-0464(26)00002-X

引用次数: 0

Augmented intelligence for multimodal virtual biopsy in breast cancer using generative artificial intelligence 基于生成人工智能的乳腺癌多模态虚拟活检增强智能。

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2025-12-26 DOI: 10.1016/j.jbi.2025.104971

Aurora Rofena , Claudia Lucia Piccolo , Bruno Beomonte Zobel , Paolo Soda , Valerio Guarrasi

Objective:

This study aims to propose a multimodal, multi-view deep learning approach for breast cancer virtual biopsy, a non-invasive classification of breast lesions as malignant or benign, by integrating Full-Field Digital Mammography (FFDM) and Contrast-Enhanced Spectral Mammography (CESM). The work addresses the critical challenge of missing CESM data by introducing generative artificial intelligence (AI) to synthesize CESM images when unavailable, ensuring the continuity of diagnostic workflows.

Methods:

The proposed method uses FFDM and CESM images in both craniocaudal (CC) and mediolateral oblique (MLO) views. When CESM is missing, a CycleGAN-based generative model produces synthetic CESM images from FFDM inputs. For classification, three convolutional neural networks (ResNet18, ResNet50, and VGG16) are employed, and a two-stage late fusion strategy integrates view-specific and modality-specific malignancy probabilities, weighted by Matthews Correlation Coefficient (MCC), into a final malignancy score. The system’s robustness under varying degrees of missing CESM data is tested by incrementally replacing real CESM inputs with synthetic ones and evaluating classification performance using AUC, G-mean, and MCC.

Results:

CycleGAN achieved high-fidelity CESM synthesis, with Peak-Signal-to-Noise Ratio exceeding 24 dB and Structural Similarity Index above 0.8 across both CC and MLO views. For lesion classification, the multimodal configuration combining FFDM and CESM consistently outperformed the unimodal FFDM-only setup. Notably, even when CESM was entirely replaced by synthetic images, the multimodal approach still improved virtual biopsy performance compared to FFDM alone. Although classification performance declined as the proportion of synthetic CESM increased, the use of synthetic data remained beneficial.

Conclusion:

This work demonstrates that generative AI can effectively address missing-modality challenges in breast cancer diagnostics by synthesizing CESM images to enhance FFDM-based virtual biopsy pipelines. In the absence of real CESM data, incorporating synthetic images improves lesion classification compared to using FFDM alone, offering a non-invasive alternative to support clinical decision-making. Moreover, by releasing the extended CESM@UCBM dataset, this study contributes a valuable resource for advancing research and innovation in breast multimodal diagnostic systems.

目的：本研究旨在通过整合全场数字乳房x线摄影（FFDM）和对比增强光谱乳房x线摄影（CESM），提出一种用于乳腺癌虚拟活检的多模式、多视图深度学习方法，对乳房病变进行恶性或良性的无创分类。这项工作通过引入生成式人工智能（AI）来合成不可用的CESM图像，从而确保诊断工作流程的连续性，解决了缺少CESM数据的关键挑战。方法：在颅侧（CC）和中外侧斜（MLO）视图上使用FFDM和CESM图像。当缺少CESM时，基于cyclegan的生成模型从FFDM输入生成合成CESM图像。为了进行分类，使用了三个卷积神经网络（ResNet18， ResNet50和VGG16），并采用两阶段后期融合策略将特定视图和特定模式的恶性肿瘤概率结合起来，通过马修斯相关系数（MCC）加权，形成最终的恶性肿瘤评分。通过逐步用合成的CESM输入替换真实的CESM输入，并使用AUC、G-mean和MCC评估分类性能，测试了系统在不同程度缺失CESM数据下的鲁棒性。结果：CycleGAN实现了高保真的CESM合成，在CC和MLO视图上，峰值信噪比超过24 dB，结构相似指数超过0.8。对于病变分类，结合FFDM和CESM的多模态配置始终优于单模态FFDM设置。值得注意的是，即使CESM完全被合成图像取代，与单独的FFDM相比，多模态方法仍然提高了虚拟活检的性能。虽然分类性能随着合成CESM比例的增加而下降，但合成数据的使用仍然是有益的。结论：本研究表明，生成式人工智能可以通过合成CESM图像来增强基于ffdm的虚拟活检管道，有效解决乳腺癌诊断中缺失模态的挑战。在缺乏真实CESM数据的情况下，与单独使用FFDM相比，结合合成图像可以改善病变分类，为支持临床决策提供非侵入性替代方案。此外，通过发布扩展的CESM@UCBM数据集，本研究为推进乳腺多模态诊断系统的研究和创新提供了宝贵的资源。

{"title":"Augmented intelligence for multimodal virtual biopsy in breast cancer using generative artificial intelligence","authors":"Aurora Rofena , Claudia Lucia Piccolo , Bruno Beomonte Zobel , Paolo Soda , Valerio Guarrasi","doi":"10.1016/j.jbi.2025.104971","DOIUrl":"10.1016/j.jbi.2025.104971","url":null,"abstract":"<div><h3>Objective:</h3><div>This study aims to propose a multimodal, multi-view deep learning approach for breast cancer virtual biopsy, a non-invasive classification of breast lesions as malignant or benign, by integrating Full-Field Digital Mammography (FFDM) and Contrast-Enhanced Spectral Mammography (CESM). The work addresses the critical challenge of missing CESM data by introducing generative artificial intelligence (AI) to synthesize CESM images when unavailable, ensuring the continuity of diagnostic workflows.</div></div><div><h3>Methods:</h3><div>The proposed method uses FFDM and CESM images in both craniocaudal (CC) and mediolateral oblique (MLO) views. When CESM is missing, a CycleGAN-based generative model produces synthetic CESM images from FFDM inputs. For classification, three convolutional neural networks (ResNet18, ResNet50, and VGG16) are employed, and a two-stage late fusion strategy integrates view-specific and modality-specific malignancy probabilities, weighted by Matthews Correlation Coefficient (MCC), into a final malignancy score. The system’s robustness under varying degrees of missing CESM data is tested by incrementally replacing real CESM inputs with synthetic ones and evaluating classification performance using AUC, G-mean, and MCC.</div></div><div><h3>Results:</h3><div>CycleGAN achieved high-fidelity CESM synthesis, with Peak-Signal-to-Noise Ratio exceeding 24 dB and Structural Similarity Index above 0.8 across both CC and MLO views. For lesion classification, the multimodal configuration combining FFDM and CESM consistently outperformed the unimodal FFDM-only setup. Notably, even when CESM was entirely replaced by synthetic images, the multimodal approach still improved virtual biopsy performance compared to FFDM alone. Although classification performance declined as the proportion of synthetic CESM increased, the use of synthetic data remained beneficial.</div></div><div><h3>Conclusion:</h3><div>This work demonstrates that generative AI can effectively address missing-modality challenges in breast cancer diagnostics by synthesizing CESM images to enhance FFDM-based virtual biopsy pipelines. In the absence of real CESM data, incorporating synthetic images improves lesion classification compared to using FFDM alone, offering a non-invasive alternative to support clinical decision-making. Moreover, by releasing the extended CESM@UCBM dataset, this study contributes a valuable resource for advancing research and innovation in breast multimodal diagnostic systems.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"174 ","pages":"Article 104971"},"PeriodicalIF":4.5,"publicationDate":"2025-12-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145850420","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}

引用次数: 0

Reviewer Acknowledgement 2025. 审稿人致谢2025。

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2025-12-25 DOI: 10.1016/j.jbi.2025.104974

引用次数: 0

Mamba-enhanced disease semantic knowledge graph for interpretable automatic ICD coding 用于可解释的自动ICD编码的mamba增强疾病语义知识图。

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2025-12-22 DOI: 10.1016/j.jbi.2025.104973

Pengli Lu , Chao Dong , Jingjin Xue , Fentang Gao

Automatic ICD coding refers to the process of using artificial intelligence methods to automatically extract information related to diseases, symptoms, diagnoses, treatments, and other relevant details from electronic health records, and convert it into codes that comply with the International Classification of Diseases (ICD) standard. Automatic ICD coding technology has been gradually improved with the advancement of deep learning, but in practical deployment, it still faces challenges such as inconsistent semantics, ambiguous labels, and limited interpretability. To address these issues, we propose a novel automatic ICD coding framework MKHCNet (Mamba-Knowledge-HPLA-ContraNorm Network) which integrates unstructured clinical knowledge representation, long-range dependency modeling, and contrastive normalization techniques to enhance coding performance. Specifically, we construct a disease semantic knowledge graph to enrich ICD label representations, employ the Mamba network to capture cross-domain dependencies, apply the ContraNorm module to enhance label separability, and propose the Hierarchical Position Label Attention (HPLA) mechanism to achieve fine-grained, attention-based interpretability. Finally, with the purpose of capturing complex nonlinear relationships more effectively and better adapting to complex patterns in EHR data, FastKAN acts as a classifier and utilizes radial basis function (RBF) for feature transformation. We conducted systematic experiments on the benchmark datasets MIMIC-FULL and MIMIC-50. The experimental results show that MKHCNet improves MaAUC and P

@

8 by 2.1% and 0.3% on MIMIC-FULL respectively compared with the best existing mainstream model. Furthermore, case studies demonstrate that the model is able to effectively identify complex semantic cues and provide strong clinical interpretability.

ICD自动编码是指利用人工智能方法，从电子健康记录中自动提取与疾病、症状、诊断、治疗等相关细节信息，并将其转换为符合国际疾病分类（ICD）标准的代码的过程。随着深度学习的推进，自动ICD编码技术逐渐得到完善，但在实际部署中，仍然面临语义不一致、标签模糊、可解释性有限等挑战。为了解决这些问题，我们提出了一种新的ICD自动编码框架MKHCNet (mamba - knowledge - hpla - contransform Network)，该框架集成了非结构化临床知识表示、远程依赖建模和对比归一化技术来提高编码性能。具体而言，我们构建了疾病语义知识图来丰富ICD标签表示，使用Mamba网络捕获跨领域依赖关系，应用contransform模块增强标签可分离性，并提出了分层位置标签注意（HPLA）机制来实现细粒度的、基于注意的可解释性。最后，为了更有效地捕捉复杂的非线性关系，更好地适应电子病历数据中的复杂模式，FastKAN作为分类器，利用径向基函数（RBF）进行特征转换。我们在基准数据集MIMIC- full和MIMIC 50上进行了系统的实验。实验结果表明，与现有最佳主流模型相比，MKHCNet在MIMIC-FULL上的MaAUC和MaF分别提高了2.2%和0.9%。此外，案例研究表明，该模型能够有效识别复杂的语义线索，并提供强大的临床可解释性。

{"title":"Mamba-enhanced disease semantic knowledge graph for interpretable automatic ICD coding","authors":"Pengli Lu , Chao Dong , Jingjin Xue , Fentang Gao","doi":"10.1016/j.jbi.2025.104973","DOIUrl":"10.1016/j.jbi.2025.104973","url":null,"abstract":"<div><div>Automatic ICD coding refers to the process of using artificial intelligence methods to automatically extract information related to diseases, symptoms, diagnoses, treatments, and other relevant details from electronic health records, and convert it into codes that comply with the International Classification of Diseases (ICD) standard. Automatic ICD coding technology has been gradually improved with the advancement of deep learning, but in practical deployment, it still faces challenges such as inconsistent semantics, ambiguous labels, and limited interpretability. To address these issues, we propose a novel automatic ICD coding framework <strong>MKHCNet</strong> (<strong>M</strong>amba-<strong>K</strong>nowledge-<strong>H</strong>PLA-<strong>C</strong>ontraNorm <strong>Net</strong>work) which integrates unstructured clinical knowledge representation, long-range dependency modeling, and contrastive normalization techniques to enhance coding performance. Specifically, we construct a disease semantic knowledge graph to enrich ICD label representations, employ the Mamba network to capture cross-domain dependencies, apply the ContraNorm module to enhance label separability, and propose the Hierarchical Position Label Attention (HPLA) mechanism to achieve fine-grained, attention-based interpretability. Finally, with the purpose of capturing complex nonlinear relationships more effectively and better adapting to complex patterns in EHR data, FastKAN acts as a classifier and utilizes radial basis function (RBF) for feature transformation. We conducted systematic experiments on the benchmark datasets MIMIC-FULL and MIMIC-50. The experimental results show that MKHCNet improves MaAUC and P<span><math><mi>@</mi></math></span>8 by 2.1% and 0.3% on MIMIC-FULL respectively compared with the best existing mainstream model. Furthermore, case studies demonstrate that the model is able to effectively identify complex semantic cues and provide strong clinical interpretability.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"173 ","pages":"Article 104973"},"PeriodicalIF":4.5,"publicationDate":"2025-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145827758","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}

引用次数: 0

Beyond manual transcripts: Exploring the potential of automatic speech recognition errors in improving Alzheimer’s disease detection 超越手工转录：探索自动语音识别错误在提高阿尔茨海默病检测中的潜力

IF 4.5 2区医学 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Journal of Biomedical Informatics

Pub Date : 2025-12-18 DOI: 10.1016/j.jbi.2025.104968

Yin-Long Liu , Yuanchao Li , Rui Feng , Jiaxin Chen , Yiming Wang , Yu-Ang Chen , Nan Ding , Jiahong Yuan , Zhen-Hua Ling

Objective:

This study aims to extend the counterintuitive observation that Automatic Speech Recognition (ASR) errors can be beneficial for Alzheimer’s Disease (AD) detection. Our objective is to conduct a large-scale investigation to validate this phenomenon and, more importantly, to elucidate the specific mechanisms by which ASR errors can serve as valuable diagnostic clues for distinguishing individuals with AD from Healthy Controls (HC).

Methods:

We employed 18 ASR models, in both their original and fine-tuned versions, to generate 36 sets of transcripts from the ADReSS dataset. We also synthesized speech from both manual and ASR transcripts using a text-to-speech (TTS) model. Knowledge-based features and pre-trained embeddings were extracted and fed into two proposed AD detection models : a self-attention model and a cross-attention-based interpretability model. To uncover the underlying mechanisms, we conducted a multi-faceted set of analyses, including examinations of ASR error types, words affected by ASR errors, linguistic comparisons, attention weight analysis, and case studies.

Results:

We demonstrate that transcripts generated by certain ASR models achieve higher AD detection accuracy than gold-standard manual transcripts. This performance gain stems not from errors in general or a high Word Error Rate (WER), but from specific and asymmetric error patterns. Our analyses reveal that these patterns amplify some pre-existing linguistic deficits in AD speech (e.g., disfluencies), thereby increasing the feature-level divergence between the AD and HC groups. Furthermore, we show that these diagnostic clues are effectively preserved when speech is synthesized from ASR transcripts, holding significant implications for data augmentation strategies in AD research.

Conclusion:

The specific, asymmetric error patterns introduced by certain ASR models enhance the distinction between AD and HC groups by amplifying pathological linguistic deficits associated with AD. This work suggests a paradigm shift for clinical ASR development: optimizing models not merely for transcription accuracy, but for their downstream diagnostic utility.

目的：本研究旨在扩展反直觉的观察，即自动语音识别（ASR）错误可能有利于阿尔茨海默病（AD）的检测。我们的目标是进行大规模的调查来验证这一现象，更重要的是，阐明ASR错误作为区分AD个体和健康对照（HC）的有价值的诊断线索的具体机制。方法：我们采用了18个原始版本和微调版本的ASR模型，从address数据集中生成36组转录本。我们还使用文本到语音（TTS）模型从手动和ASR转录本合成语音。提取基于知识的特征和预训练的嵌入，并将其输入到两种AD检测模型中：自注意模型和基于交叉注意的可解释性模型。为了揭示潜在的机制，我们进行了多方面的分析，包括检查ASR错误类型、受ASR错误影响的单词、语言比较、注意权重分析和案例研究。结果：我们证明由某些ASR模型生成的转录本比金标准人工转录本具有更高的AD检测精度。这种性能的提高不是来自一般的错误或较高的单词错误率，而是来自特定的和不对称的错误模式。我们的分析表明，这些模式放大了AD言语中一些先前存在的语言缺陷（例如，不流利），从而增加了AD和HC群体之间的特征水平差异。此外，我们发现当从ASR转录本合成语音时，这些诊断线索有效地保留了下来，这对AD研究中的数据增强策略具有重要意义。结论：某些ASR模型引入的特定的、不对称的错误模式通过放大与AD相关的病理性语言缺陷来增强AD和HC组之间的区别。这项工作提示了临床ASR发展的范式转变：优化模型不仅是为了转录准确性，而且为了它们的下游诊断效用。

{"title":"Beyond manual transcripts: Exploring the potential of automatic speech recognition errors in improving Alzheimer’s disease detection","authors":"Yin-Long Liu , Yuanchao Li , Rui Feng , Jiaxin Chen , Yiming Wang , Yu-Ang Chen , Nan Ding , Jiahong Yuan , Zhen-Hua Ling","doi":"10.1016/j.jbi.2025.104968","DOIUrl":"10.1016/j.jbi.2025.104968","url":null,"abstract":"<div><h3>Objective:</h3><div>This study aims to extend the counterintuitive observation that Automatic Speech Recognition (ASR) errors can be beneficial for Alzheimer’s Disease (AD) detection. Our objective is to conduct a large-scale investigation to validate this phenomenon and, more importantly, to elucidate the specific mechanisms by which ASR errors can serve as valuable diagnostic clues for distinguishing individuals with AD from Healthy Controls (HC).</div></div><div><h3>Methods:</h3><div>We employed 18 ASR models, in both their original and fine-tuned versions, to generate 36 sets of transcripts from the ADReSS dataset. We also synthesized speech from both manual and ASR transcripts using a text-to-speech (TTS) model. Knowledge-based features and pre-trained embeddings were extracted and fed into two proposed AD detection models : a self-attention model and a cross-attention-based interpretability model. To uncover the underlying mechanisms, we conducted a multi-faceted set of analyses, including examinations of ASR error types, words affected by ASR errors, linguistic comparisons, attention weight analysis, and case studies.</div></div><div><h3>Results:</h3><div>We demonstrate that transcripts generated by certain ASR models achieve higher AD detection accuracy than gold-standard manual transcripts. This performance gain stems not from errors in general or a high Word Error Rate (WER), but from specific and asymmetric error patterns. Our analyses reveal that these patterns amplify some pre-existing linguistic deficits in AD speech (e.g., disfluencies), thereby increasing the feature-level divergence between the AD and HC groups. Furthermore, we show that these diagnostic clues are effectively preserved when speech is synthesized from ASR transcripts, holding significant implications for data augmentation strategies in AD research.</div></div><div><h3>Conclusion:</h3><div>The specific, asymmetric error patterns introduced by certain ASR models enhance the distinction between AD and HC groups by amplifying pathological linguistic deficits associated with AD. This work suggests a paradigm shift for clinical ASR development: optimizing models not merely for transcription accuracy, but for their downstream diagnostic utility.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"173 ","pages":"Article 104968"},"PeriodicalIF":4.5,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145798054","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}

引用次数: 0