Journal of Biomedical Informatics最新文献

{"title":"Navigating regulatory challenges across the life cycle of a SaMD","authors":"Martina Francesconi ,&nbsp;Miriam Cangi ,&nbsp;Silvia Tamarri ,&nbsp;Noemi Conditi ,&nbsp;Chiara Menicucci ,&nbsp;Alice Ravizza ,&nbsp;Luisa Cattaneo ,&nbsp;Elisabetta Bianchini","doi":"10.1016/j.jbi.2025.104856","DOIUrl":"10.1016/j.jbi.2025.104856","url":null,"abstract":"<div><h3>Objective</h3><div>Software as medical devices (SaMDs) have become part of clinical practice and the management of the development and control processes of the documentation associated with them are an integral part of many medical realities. The European Regulation, MDR (EU) 2017/745, introduces a classification rule (rule 11, Annex VIII) specifically for software, which provides more explicit requirements than in the past, leading to classification of many software to higher risk and therefore to more complex certification processes. In this context, planning and awareness of possible regulatory strategies and related standards are fundamental for the key stakeholders, but this complex landscape can be perceived as fragmented. The aim of this work is to provide an amalgamated overview of how the current EU normative framework integrates into the various phases of the life-cycle of a medical device software, trying to ensure its safe and effective development.</div></div><div><h3>Methods</h3><div>In addition to the MDR, the main normative references relevant to the medical device software sector were taken into consideration. Specifically, the IEC 62304 standard clarifies the main processes of the software life-cycle, including the analysis of problems and changes, and the IEC 82304 standard completes its management by addressing activities relating to post-market phases and requirements. In addition, the various steps include also key points such as risk identification and control (ISO 14971), design, implementation and validation of usability requirements (IEC 62366) and in general the quality of the context in which the software is developed and maintained (ISO 13485). The application of these standards can support the activities of the various stakeholders and facilitate evidence of compliance with the regulatory requirements by MDR.</div></div><div><h3>Results</h3><div>Based on the software life cycle, a mapping of the requirements from the entire normative framework analyzed over the various phases was implemented.</div></div><div><h3>Conclusions</h3><div>A detailed and integrated picture of the regulatory context behind the life cycle of a SaMD has been provided: this can facilitate the implementation of a balanced and effective approach, including key aspects, such as risk management and usability processes, and ensuring safety for the end user.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"167 ","pages":"Article 104856"},"PeriodicalIF":4.0,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144131252","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":"Impacts of sample weighting on transferability of risk prediction models across EHR-Linked biobanks with different recruitment strategies","authors":"Maxwell Salvatore ,&nbsp;Alison M Mondul ,&nbsp;Christopher R Friese ,&nbsp;David Hanauer ,&nbsp;Hua Xu ,&nbsp;Celeste Leigh Pearce ,&nbsp;Bhramar Mukherjee","doi":"10.1016/j.jbi.2025.104853","DOIUrl":"10.1016/j.jbi.2025.104853","url":null,"abstract":"<div><h3>Objective</h3><div>To evaluate whether using poststratification weights when training risk prediction models enhances transferability when the external test cohort has a different sampling strategy, a commonly encountered scenario when analyzing electronic health record (EHR)-linked biobanks.</div></div><div><h3>Methods</h3><div>PS weights were calculated to align a health system-based biobank, the Michigan Genomics Initiative (MGI; n = 76,757), with a nationally recruited biobank, All of Us (AOU; n = 226,764), which oversamples underrepresented groups. Basic PS weights (PS_BASIC) captured age, sex, and race/ethnicity; full PS weights (PS_FULL) additionally included smoking, alcohol consumption, BMI, depression, hypertension, and the Charlson Comorbidity Index. Models for esophageal, liver, and pancreatic cancers were developed using EHR data from MGI at 0, 1, 2, and 5 years prior to diagnosis. Phenotype risk scores (PheRS) were constructed using six methods (e.g., regularized regression, random forest) and evaluated alongside covariates, risk factors, and symptoms. Evaluation metrics included the odds ratio (OR) for the top decile vs. the middle 40th-60th percentiles of the risk score distribution and the area under the receiver operating curve (AUC) evaluated in the AOU test cohort when models are trained with and without weighting.</div></div><div><h3>Results</h3><div>Elastic net and random forest methods generally performed well in risk stratification, but no single PheRS construction method consistently outperformed others. Applying PS weights did not consistently improve risk stratification performance. For example, in liver cancer risk stratification at t = 1, unweighted random forest PheRS yielded an OR of 13.73 (95 % CI: 8.97, 21.01), compared to 14.55 (95 % CI: 9.45, 22.42) with PS_BASIC and 13.62 (95 % CI: 8.90, 20.85) with PS_FULL.</div></div><div><h3>Conclusion</h3><div>PS weights do not significantly enhance risk model transferability between biobanks. EHR-based PheRS are crucial for risk stratification and should be integrated with other multimodal data for improved risk prediction. Identifying high-risk populations for diseases like liver cancer early through health history mining shows promise.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"167 ","pages":"Article 104853"},"PeriodicalIF":4.0,"publicationDate":"2025-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144116244","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":"Adaptive debiasing learning for drug repositioning","authors":"Yajie Meng ,&nbsp;Yi Wang ,&nbsp;Xinrong Hu ,&nbsp;Changcheng Lu ,&nbsp;Xianfang Tang ,&nbsp;Feifei Cui ,&nbsp;Pan Zeng ,&nbsp;Yuhua Yao ,&nbsp;Jialiang Yang ,&nbsp;Junlin Xu","doi":"10.1016/j.jbi.2025.104843","DOIUrl":"10.1016/j.jbi.2025.104843","url":null,"abstract":"<div><div>Drug repositioning, pivotal in current pharmaceutical development, aims to find new uses for existing drugs, offering an efficient and cost-effective path to drug discovery. In recent years, graph neural network-based deep learning methods have achieved significant success in drug repositioning tasks. However, few studies have analyzed the characteristics of datasets to mitigate potential data biases. In this paper, we analyzed three commonly used drug repositioning datasets and identified a consistent characteristic among them: a trend of node polarization, characterized by the presence of popular entities (those commonly occurring and extensively associated) and long-tail entities (those appearing less frequently with fewer associations). Based on this finding, we propose a deep learning framework with a debiasing mechanism, called DRDM. The framework excels in addressing popular entities’ biases, which often overshadow the subtle patterns in long-tail entities—key for novel insights. DRDM dynamically adjusts association weights during training, enhancing long-tail entity representation and reducing bias. In addition, we employ dual-view contrastive learning to provide rich supervisory signals, thereby further enhancing the model’s robustness. We conducted experiments with our method on these three datasets, and the results demonstrated that our approach exhibits strong competitiveness compared to competing models. Case studies further highlighted the potential of the model in practical applications, which could provide new insights for future drug discovery.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"167 ","pages":"Article 104843"},"PeriodicalIF":4.0,"publicationDate":"2025-05-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144071857","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":"Biomedical text normalization through generative modeling","authors":"Jacob S. Berkowitz,&nbsp;Apoorva Srinivasan,&nbsp;Jose Miguel Acitores Cortina,&nbsp;Yasaman Fatapour,&nbsp;Nicholas P Tatonetti","doi":"10.1016/j.jbi.2025.104850","DOIUrl":"10.1016/j.jbi.2025.104850","url":null,"abstract":"<div><h3>Objective</h3><div>A large proportion of electronic health record (EHR) data consists of unstructured medical language text. The formatting of this text is often flexible and inconsistent, making it challenging to use for predictive modeling, clinical decision support, and data mining. Large language models’ (LLMs) ability to understand context and semantic variations makes them promising tools for standardizing medical text. In this study, we develop and assess clinical text normalization pipelines built using large-language models.</div></div><div><h3>Methods</h3><div>We implemented four LLM-based normalization strategies (Zero-Shot Recall, Prompt Recall, Semantic Search, and Retrieval-Augmented Generation based normalization [RAGnorm]) and one baseline approach using TF-IDF based String Matching. We evaluated performance across three datasets of SNOMED-mapped condition terms: [<span><span>1</span></span>] an oncology-specific dataset, [<span><span>2</span></span>] a representative sample of institutional medical conditions, and [<span><span>3</span></span>] a dataset of commonly occurring condition codes (&gt;1000 uses) from our institution. We measured performance by recording the mean shortest path length between predicted and true SNOMED CT terms. Additionally, we benchmarked our models against the TAC 2017 drug label annotations, which normalizes terms to the Medical Dictionary for Regulatory Activities (MedDRA) Preferred Terms.</div></div><div><h3>Results</h3><div>We found that RAGnorm was the most effective throughout each dataset, achieving a mean shortest path length of 0.21 for the domain-specific dataset, 0.58 for the sampled dataset, and 0.90 for the top terms dataset. It achieved a micro F1 score of 88.01 on task 4 of the TAC2017 conference, surpassing all other models without viewing the provided training data.</div></div><div><h3>Conclusion</h3><div>We find that retrieval-focused approaches overcome traditional LLM limitations for this task. RAGnorm and related retrieval techniques should be explored further for the normalization of biomedical free text.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"167 ","pages":"Article 104850"},"PeriodicalIF":4.0,"publicationDate":"2025-05-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144093797","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":"Machine learning applications related to suicide in military and Veterans: A scoping literature review","authors":"Yuhan Zhang ,&nbsp;Yishu Wei ,&nbsp;Yanshan Wang ,&nbsp;Yunyu Xiao ,&nbsp;COL Ret. Ronald K. Poropatich ,&nbsp;Gretchen L. Haas ,&nbsp;Yiye Zhang ,&nbsp;Chunhua Weng ,&nbsp;Jinze Liu ,&nbsp;Lisa A. Brenner ,&nbsp;James M. Bjork ,&nbsp;Yifan Peng","doi":"10.1016/j.jbi.2025.104848","DOIUrl":"10.1016/j.jbi.2025.104848","url":null,"abstract":"<div><h3>Objective</h3><div>Suicide remains one of the main preventable causes of death among service members and veterans. Early detection and accurate prediction are essential components of effective suicide prevention strategies. Machine learning techniques have been explored in recent years with a specific focus on the assessment and prediction of multiple suicide-related outcomes, showing promising advancements. This study aims to assess and summarize current research and provides a comprehensive review regarding the application of machine learning techniques in assessing and predicting suicidal ideation, attempts, and mortality among members of military and veteran populations.</div></div><div><h3>Methods</h3><div>A keyword search using PubMed, IEEE, ACM, and Google Scholar was conducted, and the PRISMA protocol was adopted for relevant study selection. Peer-reviewed original research in English targeting the assessment or prediction of suicide-related outcomes among service members and veteran populations was included. 1,110 studies were retrieved, and 32 satisfied the inclusion criteria and were included.</div></div><div><h3>Results</h3><div>Thirty-two articles met the inclusion criteria. Despite these studies exhibiting significant variability in sample characteristics, data modalities, specific suicide-related outcomes, and the machine learning technologies employed, they consistently identified risk factors relevant to mental health issues such as depression, post-traumatic stress disorder (PTSD), suicidal ideation, prior attempts, physical health problems, and demographic characteristics. Machine learning models applied in this area have demonstrated reasonable predictive accuracy and have verified, on a large scale, risk factors previously detected by more manual analytic methods. Additional research gaps still exist. First, many studies have overlooked metrics that distinguish between false positives and negatives, such as positive predictive value and negative predictive value, which are crucial in the context of suicide prevention policies. Second, more dedicated approaches to handling survival and longitudinal data should be explored. Lastly, most studies focused on machine learning methods, with limited discussion of their connection to clinical rationales.</div></div><div><h3>Conclusion</h3><div>In sum, machine learning analyses have identified risk factors associated with suicide in military populations, which span a wide range of psychological, biological, and sociocultural factors, highlighting the complexities involved in assessing suicide risk among service members and veterans. Some differences were noted between males and females. The diversity of these factors also demonstrates that effective prevention strategies must be comprehensive and flexible.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"167 ","pages":"Article 104848"},"PeriodicalIF":4.0,"publicationDate":"2025-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144078291","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":"Multi-agent norm perception and induction in distributed healthcare","authors":"Chao Li ,&nbsp;Olga Petruchik ,&nbsp;Elizaveta Grishanina ,&nbsp;Sergey Kovalchuk","doi":"10.1016/j.jbi.2025.104835","DOIUrl":"10.1016/j.jbi.2025.104835","url":null,"abstract":"<div><div>This paper presents a Multi-Agent Norm Perception and Induction Learning Model aimed at facilitating the integration of autonomous agent systems into distributed healthcare environments through dynamic interaction processes. The nature of the medical norm system and its sharing channels necessitates distinct approaches for Multi-Agent Systems to learn two types of norms. Building on this foundation, the model enables agents to simultaneously learn descriptive norms, which capture collective tendencies, and prescriptive norms, which dictate ideal behaviors. Through parameterized mixed probability density models and practice-enhanced Markov games, the multi-agent system perceives descriptive norms in dynamic interactions and captures emergent prescriptive norms. We conducted experiments using a dataset from a neurological medical center spanning from 2016 to 2020.</div><div>The descriptive norm-sharing experiment results demonstrate that the model can effectively perceive the descriptive collective medical norms – which embody the current best clinical practices – across medical communities of varying scales. By contrasting this with the fact that the real descriptive diagnostic practice patterns in the neurological medical center dataset gradually converged over a period of 5 years, we find that the model, through prolonged learning and sharing processes, progressively mirrors the actual descriptive diagnostic trends and collective behavioral tendencies present within the medical community. In the experiment where multiple agents infer prescriptive norms within a dynamic healthcare environment, the agents effectively learned the key clinical protocols within the norm space <span><math><mi>H</mi></math></span>, which includes control norms, without developing high belief in invalid norms. Furthermore, the agents’ belief update process was relatively smooth, avoiding any discontinuous stepwise updates.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"166 ","pages":"Article 104835"},"PeriodicalIF":4.0,"publicationDate":"2025-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143984923","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":"A lightweight graph neural network to predict long-term mortality in coronary artery disease patients: an interpretable causality-aware approach","authors":"Mohammad Yaseliani ,&nbsp;Md. Noor-E-Alam ,&nbsp;Osama Dasa ,&nbsp;Xiaochen Xian ,&nbsp;Carl J. Pepine ,&nbsp;Md Mahmudul Hasan","doi":"10.1016/j.jbi.2025.104846","DOIUrl":"10.1016/j.jbi.2025.104846","url":null,"abstract":"<div><h3>Background</h3><div>Coronary artery disease (CAD) causes substantial death toll in the United States and worldwide. While traditional methods for CAD mortality prediction are based on established risk factors, they have significant limitations in accuracy, adaptability to diverse populations, performance for individual risk prediction compared to group data, and incorporation of socioeconomic and lifestyle variations. Machine learning (ML) models have demonstrated superior performance in CAD prediction; however, they often struggle with capturing complex data interactions that can impact mortality.</div></div><div><h3>Methods</h3><div>We proposed lightweight, interpretable graph neural network (GNN) models, utilizing data from a large trial of hypertensive patients with CAD to predict mortality using a concise set of critical features. While this smaller set of features can improve efficiency and implementation in clinical settings, the model’s “lightweight” nature facilitates fast real-time applications. We utilized a hybrid approach, which first uses logistic regression (LR) to identify statistically significant features, followed by propensity score matching (PSM) to identify potentially causal features. These causal features, alongside demographic variables, were employed to create a graph of patients, drawing edges between patients with similar causal features. Accordingly, lightweight 5-layer graph convolutional network (GCN) and graph attention network (GAT) were designed for mortality prediction, followed by an interpretable method (i.e., GNNExplainer) to report the feature importance.</div></div><div><h3>Results</h3><div>The proposed GCN achieved a recall of 93.02% and a negative predictive value (NPV) of 89.42%, higher than all other classifiers. Accordingly, a web-based decision support system (DSS), called CAD-SS, was developed, capable of predicting mortality and identifying risk factors and similar patients, guiding clinicians in reliable and informed decision-making.</div></div><div><h3>Conclusions</h3><div>Our proposed CAD-SS, which utilizes an interpretable and causality-aware lightweight GCN model, demonstrated reasonably high performance in predicting mortality due to CAD. This unique system can help identify the most vulnerable patients.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"167 ","pages":"Article 104846"},"PeriodicalIF":4.0,"publicationDate":"2025-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144016306","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":"Evaluating an information theoretic approach for selecting multimodal data fusion methods","authors":"Tengyue Zhang ,&nbsp;Ruiwen Ding ,&nbsp;Kha-Dinh Luong ,&nbsp;William Hsu","doi":"10.1016/j.jbi.2025.104833","DOIUrl":"10.1016/j.jbi.2025.104833","url":null,"abstract":"<div><h3>Objective:</h3><div>Interest has grown in combining radiology, pathology, genomic, and clinical data to improve the accuracy of diagnostic and prognostic predictions toward precision health. However, most existing works choose their datasets and modeling approaches empirically and in an ad hoc manner. A prior study proposed four partial information decomposition (PID)-based metrics to provide a theoretical understanding of multimodal data interactions: redundancy, uniqueness of each modality, and synergy. However, these metrics have only been evaluated in a limited collection of biomedical data, and the existing work does not elucidate the effect of parameter selection when calculating the PID metrics. In this work, we evaluate PID metrics on a wider range of biomedical data, including clinical, radiology, pathology, and genomic data, and propose potential improvements to the PID metrics.</div></div><div><h3>Methods:</h3><div>We apply the PID metrics to seven different modality pairs across four distinct cohorts (datasets). We compare and interpret trends in the resulting PID metrics and downstream model performance in these multimodal cohorts. The downstream tasks being evaluated include predicting the prognosis (either overall survival or recurrence) of patients with non-small cell lung cancer, prostate cancer, and glioblastoma.</div></div><div><h3>Results:</h3><div>We found that, while PID metrics are informative, solely relying on these metrics to decide on a fusion approach does not always yield a machine learning model with optimal performance. Of the seven different modality pairs, three had poor (0%), three had moderate (66%–89%), and only one had perfect (100%) consistency between the PID values and model performance. We propose two improvements to the PID metrics (determining the optimal parameters and uncertainty estimation) and identified areas where PID metrics could be further improved.</div></div><div><h3>Conclusion:</h3><div>The current PID metrics are not accurate enough for estimating the multimodal data interactions and need to be improved before they can serve as a reliable tool. We propose improvements and provide suggestions for future work. Code: <span><span>https://github.com/zhtyolivia/pid-multimodal</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"167 ","pages":"Article 104833"},"PeriodicalIF":4.0,"publicationDate":"2025-05-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144020795","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":"Knowledge-enhanced Parameter-efficient Transfer Learning with METER for medical vision-language tasks","authors":"Xudong Liang ,&nbsp;Jiang Xie ,&nbsp;Jinzhu Wei ,&nbsp;Mengfei Zhang ,&nbsp;Haoyang Zhang","doi":"10.1016/j.jbi.2025.104840","DOIUrl":"10.1016/j.jbi.2025.104840","url":null,"abstract":"<div><h3>Objective:</h3><div>The full fine-tuning paradigm becomes impractical when applying pre-trained models to downstream tasks due to significant computational and storage costs. Parameter-efficient fine-tuning (PEFT) methods can alleviate the issue. However, solely applying PEFT methods leads to sub-optimal performance owing to the domain gap between pre-trained models and medical downstream tasks.</div></div><div><h3>Methods:</h3><div>This study proposes <u>K</u>nowledge-enhanced <u>P</u>arameter-efficient Transfer <u>L</u>earning with <u>METER</u> (KPL-METER) for medical vision-language (VL) downstream tasks. KPL-METER combines PEFT methods, including an innovative PEFT module for multi-modal branches and newly introduced external domain-specific knowledge to enhance model performance. First, a lightweight, plug-and-play module named Sharing Adapter (SAdapter) is developed and inserted into the multi-modal encoders. This allows the two modalities to maintain uni-modal features while encouraging cross-modal consistency. Second, a novel knowledge extraction method and a parameter-free knowledge modeling strategy are developed to incorporate domain-specific knowledge from the Unified Medical Language System (UMLS) into multi-modal features. To further enhance the modeling of uni-modal features, Adapter is added to the image and text encoders.</div></div><div><h3>Results:</h3><div>The effectiveness of the proposed model is evaluated on two medical VL tasks using three VL datasets. The results indicate that the KPL-METER model outperforms other PEFT methods in terms of performance while utilizing fewer parameters. Furthermore, KPL-METER-MED, which incorporates medical-tailored encoders, is developed. Compared to previous models in the medical domain, KPL-METER-MED tunes fewer parameters while generally achieving higher performance.</div></div><div><h3>Conclusion:</h3><div>The proposed KPL-METER architecture effectively adapts general VL models for medical VL tasks, and the designed knowledge extraction and fusion method notably enhance performance by integrating medical domain-specific knowledge. Code is available at <span><span>https://github.com/Adam-lxd/KPL-METER</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"166 ","pages":"Article 104840"},"PeriodicalIF":4.0,"publicationDate":"2025-05-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143947056","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":"Implementation of a Digital Maturity Framework for Biobanking","authors":"Federica Rossi ,&nbsp;Davide Fragnito ,&nbsp;Antonella Cruoglio ,&nbsp;Ramona Palombo ,&nbsp;Alice Massacci ,&nbsp;Alessandro Sulis ,&nbsp;Vittorio Meloni ,&nbsp;Sara Casati ,&nbsp;Antonella Mirabile ,&nbsp;Andrea Manconi ,&nbsp;Luciano Milanesi ,&nbsp;Gennaro Ciliberto ,&nbsp;Monica Forni ,&nbsp;Valentina Adami ,&nbsp;Massimiliano Borsani ,&nbsp;Claudia Miele ,&nbsp;Marialuisa Lavitrano ,&nbsp;Matteo Pallocca","doi":"10.1016/j.jbi.2025.104842","DOIUrl":"10.1016/j.jbi.2025.104842","url":null,"abstract":"<div><h3>Objective</h3><div>Digitalization is a pillar of reproducible research and a mandatory requirement for Research Infrastructures. Biobanks must ensure a fully engineered and digitalized process towards data FAIRification. To this aim, the first step is to assess the current level of digitalization using quantitative metrics, which is particularly challenging given the multi-faceted regulatory and logistical nature of biobanking.</div></div><div><h3>Methods</h3><div>We developed a Biobanking digital assessment maturity framework, BB4FAIR, comprising a survey divided into three macro areas, namely IT infrastructure, personnel, and data annotation richness. Furthermore, we implemented an automated R/Shiny system to analyse survey responses and generate visual data representations. We piloted the tool on 46 Italian biobanks that in 2023 had signed the partner charter with BBMRI. A scoring table facilitated the tiering of digital maturity, highlighting areas requiring corrective action.</div></div><div><h3>Results</h3><div>The assessment revealed significant heterogeneity across the three macro-areas of digitalization: almost half of the biobanks feature adequate IT infrastructure and personnel, and a smaller proportion have robust data annotation capabilities. Notably, most biobanks reported having a Biobank IT Management System (BIMS) or an alternative that serves their purposes, yet they still collect the consent to biobanking for future purposes in paper format; the digitalization of informed consent is generally lacking. These findings highlight the need for targeted improvements in Biobank digitalization to enhance overall data FAIRness.</div></div><div><h3>Conclusion</h3><div>The survey results underscore a pressing need for enhanced IT training and improved data annotation resources within the BBMRI.it. Corrective actions on many lacking features and desiderata are ongoing in the context of the #NextGenerationEu “Strengthening BBMRI.it” project.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"166 ","pages":"Article 104842"},"PeriodicalIF":4.0,"publicationDate":"2025-05-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144012739","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}