Artificial Intelligence in Medicine最新文献

{"title":"Position of artificial intelligence in healthcare and future perspective","authors":"Vedat Cicek,&nbsp;Ulas Bagci","doi":"10.1016/j.artmed.2025.103193","DOIUrl":"10.1016/j.artmed.2025.103193","url":null,"abstract":"<div><div>Artificial Intelligence (AI) has been used in healthcare with increasing momentum. According to a published report, 6.6 billion dollars were invested in AI healthcare in 2021, and this investment is expected to provide 150 billion dollars of benefit to the USA economy by 2026 (Duchateau and King, 2023 [1]). The future perspective on AI will undoubtedly open new horizons for the healthcare.</div><div>AI technology in the healthcare field is increasingly popular in the areas of diagnosis, prognosis, classification, therapy, and disease survival prediction. Now that AI has proven its worth, it's already time to re-ask the following three questions according to the fast pace of AI algorithms:<ul><li><span>1)</span><span><div>Where will AI be positioned in healthcare in the future?</div></span></li><li><span>2)</span><span><div>What kind of relationship will be defined between doctors, patients and AI?</div></span></li><li><span>3)</span><span><div>How can we direct AI studies according to the health problems in the world?</div></span></li></ul></div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103193"},"PeriodicalIF":6.1,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144270666","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":"Toward responsible artificial intelligence in medicine: Reflections from the Australian epilepsy project","authors":"Mangor Pedersen ,&nbsp;Heath R. Pardoe ,&nbsp;Anton de Weger ,&nbsp;Donna Hutchison ,&nbsp;David F. Abbott ,&nbsp;Karin Verspoor ,&nbsp;Graeme D. Jackson ,&nbsp;for the Australian Epilepsy Project Investigators","doi":"10.1016/j.artmed.2025.103192","DOIUrl":"10.1016/j.artmed.2025.103192","url":null,"abstract":"<div><div>Artificial intelligence (AI) is a multidisciplinary scientific field that uses machines to solve real-world problems and predict outcomes. Despite the current enthusiasm about AI's potential as a clinical support tool, there is also a growing awareness and concern about the potentially harmful effects of AI. Because AI will likely impact expert-based decision-making in medicine, it is critical to consider the issues that AI raises in medical research. This paper outlines the AI guidelines of the Australian Epilepsy Project. This large-scale platform aims to democratise specialist care in epilepsy and use AI for clinical decision support based on prospective multimodal datasets (MRI, genetic, clinical, and cognitive data) from thousands of people with epilepsy. As AI develops rapidly, we focus on key areas of medical AI identified in the literature, including <em>Trust, Responsibility</em> and <em>Safety.</em> We believe AI is changing medicine, and we believe it is imperative to advance and update our AI guidelines adaptably while preparing for an era of augmented-intelligence-based medicine.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103192"},"PeriodicalIF":6.1,"publicationDate":"2025-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144307328","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":"Lesion boundary detection for skin lesion segmentation based on boundary sensing and CNN-transformer fusion networks","authors":"Xuzhen Huang ,&nbsp;Yuliang Ma ,&nbsp;Xiajin Mei ,&nbsp;Zizhuo Wu ,&nbsp;Mingxu Sun ,&nbsp;Qingshan She","doi":"10.1016/j.artmed.2025.103190","DOIUrl":"10.1016/j.artmed.2025.103190","url":null,"abstract":"<div><div>Traditional convolutional neural networks often struggle to capture global information and handle ambiguous boundaries during complex skin lesion segmentation tasks. To tackle this challenge, we proposed MPBA-Net, a hybrid network that integrates multi-pooling fusion and boundary-aware refinement. The network integrated Convolutional Neural Network (CNN) and Transformer to generate rich skin lesion feature maps for comprehensive feature extraction. Specifically, we introduced a boundary-aware attention gate (BAAG) module in the Transformer encoder layer and added a boundary cross attention (BCA) module at the end of the network to capture critical skin lesion boundary features. Additionally, we developed a multi-pooling fusion (MPF) module that extracts global multi-scale features by fusing improved Spatial Pyramid (SP) and Atrous Spatial Pyramid Pooling (ASPP). To optimize training, we designed a Point Loss derived from Binary Cross-Entropy (BCE) and combined it with Dice Loss to form a hybrid loss function. This approach not only enhances classification performance but also provides more precise measurement of the similarity between segmentation results and ground truth annotations. Ablation experiments on the ISIC2018 dataset validated the effectiveness of our fusion strategies and network improvements. Comparative experiments on the ISIC2016, ISIC2017, and ISIC2018 datasets showed that the Dice index of MPBA-Net outperformed other comparative segmentation methods in all three datasets, achieving 91.47 %, 87.04 %, and 88.93 %, respectively. Quantitative and qualitative results demonstrate that our method improves skin lesion segmentation accuracy, aiding dermatologists in clinical diagnosis and treatment. Our code is available at <span><span>https://github.com/FengYuchenGuang/MPBA-Net</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103190"},"PeriodicalIF":6.1,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144223034","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":"Combining structural equation modeling analysis with machine learning for early malignancy detection in Bethesda Category III thyroid nodules","authors":"Zeliha Aydın Kasap ,&nbsp;Burçin Kurt ,&nbsp;Ali Güner ,&nbsp;Elif Özsağır ,&nbsp;Mustafa Emre Ercin","doi":"10.1016/j.artmed.2025.103186","DOIUrl":"10.1016/j.artmed.2025.103186","url":null,"abstract":"<div><div>Atypia of Undetermined Significance (AUS), classified as Category III in the Bethesda Thyroid Cytopathology Reporting System, presents significant diagnostic challenges for clinicians. This study aims to develop a clinical decision support system that integrates structural equation modeling (SEM) and machine learning to predict malignancy in AUS thyroid nodules. The model integrates preoperative clinical data, ultrasonography (USG) findings, and cytopathological and morphometric variables.</div><div>This retrospective cohort study was conducted between 2011 and 2019 at Karadeniz Technical University (KTU) Farabi Hospital. The dataset included 56 variables derived from 204 thyroid nodules diagnosed via ultrasound-guided fine-needle aspiration biopsy (FNAB) in 183 patients over 18 years. Logistic regression (LR) and SEM were used to identify risk factors for early thyroid cancer detection. Subsequently, machine learning algorithms—including Support Vector Machines (SVM), Naive Bayes (NB), and Decision Trees (DT) were used to construct decision support models.</div><div>After feature selection with SEM, the SVM model achieved the highest performance, with an accuracy of 82 %, a specificity of 97 %, and an AUC value of 84 %. Additional models were developed for different scenarios, and their performance metrics were compared. Accurate preoperative prediction of malignancy in thyroid nodules is crucial for avoiding unnecessary surgeries. The proposed model supports more informed clinical decision-making by effectively identifying benign cases, thereby reducing surgical risk and improving patient care.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103186"},"PeriodicalIF":6.1,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144205560","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 cell-interacting and multi-correcting method for automatic circulating tumor cells detection","authors":"Xuan Zhang ,&nbsp;Rensheng Lai ,&nbsp;Ling Bai ,&nbsp;Jianxin Ji ,&nbsp;Ruihao Qin ,&nbsp;Lihong Jiang ,&nbsp;Bin Meng ,&nbsp;Ying Zhang ,&nbsp;Xiaohan Zheng ,&nbsp;Yan Wang ,&nbsp;Xiang Kui ,&nbsp;Liuchao Zhang ,&nbsp;Dimin Ning ,&nbsp;Liuying Wang ,&nbsp;Yujiang Chen ,&nbsp;Xinling Wang ,&nbsp;Shuang Li ,&nbsp;Menglei Hua ,&nbsp;Junkai Wang ,&nbsp;Yong Cao ,&nbsp;Lei Cao","doi":"10.1016/j.artmed.2025.103164","DOIUrl":"10.1016/j.artmed.2025.103164","url":null,"abstract":"<div><div>Sensitive detection of circulating tumor cells (CTCs) from peripheral blood can serve as an effective tool in the early diagnosis and prognosis of cancer. Many methods based on modern object detectors were proposed in recent years for automatic abnormal cells detection in slide images. Although the modes of these methods can also be applied to the CTCs detection, several practical difficulties lead to suboptimal performance of them, such as accurate capture of CTCs in a large number of mixed cells and identification of CTCs and CTC-like cells with similar visual characteristics. Here, we develop a new cell-interacting and multi-correcting detector called CMD, and apply H&amp;E-stained slide images to detect CTCs automatically for the first time. Specifically, the proposed method incorporates two task-oriented novel modules: (1) a self-attention module for aggregating feature interactions between cells and allowing the model to pay more attention to key abnormal cells, (2) a hard sample mining sampler for progressively correcting predictions of cells with ambiguous classification boundaries. Experiments conducted on a multi-center dataset of 1247 annotated slide images confirm the superiority of our method over state-of-the-art cell detection methods. The results of ablation experiment part also prove the effectiveness of two modules. The source codes of this paper are available at <span><span>https://github.com/zx333445/CMD</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103164"},"PeriodicalIF":6.1,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169170","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":"PhenoLinker: Phenotype-gene link prediction and explanation using heterogeneous graph neural networks","authors":"Jose L. Mellina Andreu ,&nbsp;Luis Bernal ,&nbsp;Antonio F. Skarmeta ,&nbsp;Mina Ryten ,&nbsp;Sara Álvarez ,&nbsp;Alejandro Cisterna García ,&nbsp;Juan A. Botía","doi":"10.1016/j.artmed.2025.103177","DOIUrl":"10.1016/j.artmed.2025.103177","url":null,"abstract":"<div><div>The association of a given human phenotype with a genetic variant remains a critical challenge in biomedical research. We present PhenoLinker, a novel graph-based system capable of associating a score to a phenotype-gene relationship by using heterogeneous information networks and a convolutional neural network-based model for graphs, which can provide an explanation for the predictions. Unlike previous approaches, PhenoLinker integrates gene and phenotype attributes, while maintaining explainability through Integrated Gradients. PhenoLinker consistently outperforms existing models in both retrospective and temporal validation tasks. This system can aid in the discovery of new associations and in understanding the consequences of human genetic variation.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103177"},"PeriodicalIF":6.1,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169167","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":"An AI system for continuous knee osteoarthritis severity grading: An anomaly detection inspired approach with few labels","authors":"Niamh Belton ,&nbsp;Aonghus Lawlor ,&nbsp;Kathleen M. Curran","doi":"10.1016/j.artmed.2025.103138","DOIUrl":"10.1016/j.artmed.2025.103138","url":null,"abstract":"<div><div>The diagnostic accuracy and subjectivity of existing Knee Osteoarthritis (OA) ordinal grading systems has been a subject of on-going debate and concern. Existing automated solutions are trained to emulate these imperfect systems, whilst also being reliant on large annotated databases for fully-supervised training. This work proposes a three stage approach for automated continuous grading of knee OA that is built upon the principles of Anomaly Detection (AD); learning a robust representation of healthy knee X-rays and grading disease severity based on its distance to the centre of normality. In the first stage, SS-FewSOME is proposed, a self-supervised AD technique that learns the ‘normal’ representation, requiring only examples of healthy subjects and <span><math><mrow><mo>&lt;</mo><mn>3</mn><mtext>%</mtext></mrow></math></span> of the labels that existing methods require. In the second stage, this model is used to pseudo label a subset of unlabelled data as ‘normal’ or ‘anomalous’, followed by denoising of pseudo labels with CLIP. The final stage involves retraining on labelled and pseudo labelled data using the proposed Dual Centre Representation Learning (DCRL) which learns the centres of two representation spaces; normal and anomalous. Disease severity is then graded based on the distance to the learned centres. The proposed methodology outperforms existing techniques by margins of up to 24% in terms of OA detection and the disease severity scores correlate with the Kellgren-Lawrence grading system at the same level as human expert performance. Code available at <span><span>https://github.com/niamhbelton/SS-FewSOME_Disease_Severity_Knee_Osteoarthritis</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103138"},"PeriodicalIF":6.1,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169169","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":"Large scale gene set ranking for survival-related gene sets","authors":"Martin Špendl ,&nbsp;Jaka Kokošar ,&nbsp;Ela Praznik ,&nbsp;Luka Ausec ,&nbsp;Miha Štajdohar ,&nbsp;Blaž Zupan","doi":"10.1016/j.artmed.2025.103149","DOIUrl":"10.1016/j.artmed.2025.103149","url":null,"abstract":"<div><div>Disease progression is closely linked to shifts in the expression levels of specific genes within molecular pathways. While gene set enrichment analysis is a widely employed method for identifying key disease markers, it has been underutilized in survival analysis. Here, we introduce a novel computational approach that adapts gene set enrichment analysis for survival analysis. The proposed approach considers a gene set, computes a single-sample gene set enrichment score, and, based on this score, splits the samples into cohorts. It then scores the gene sets by evaluating the differences in survival rates between the resulting cohorts. We aim to find gene sets that can lead to cohorts with significantly different survival probabilities. Utilizing gene expression data from The Cancer Genome Atlas and gene sets from the Molecular Signature Database, our results demonstrate that existing empirical research consistently supports the top gene sets our approach associates with survival prognosis. The proposed method broadens gene set enrichment analysis applications to include information on survival, bridging the gap between alterations in molecular pathways and their implications on survival.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103149"},"PeriodicalIF":6.1,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169825","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":"Quantitative computed tomography imaging classification of cement dust-exposed patients-based Kolmogorov-Arnold networks","authors":"Ngan-Khanh Chau ,&nbsp;Woo Jin Kim ,&nbsp;Chang Hyun Lee ,&nbsp;Kum Ju Chae ,&nbsp;Gong Yong Jin ,&nbsp;Sanghun Choi","doi":"10.1016/j.artmed.2025.103166","DOIUrl":"10.1016/j.artmed.2025.103166","url":null,"abstract":"<div><h3>Background</h3><div>Occupational health assessment is critical for detecting respiratory issues caused by harmful exposures, such as cement dust. Quantitative computed tomography (QCT) imaging provides detailed insights into lung structure and function, enhancing the diagnosis of lung diseases. However, its high dimensionality poses challenges for traditional machine learning methods.</div></div><div><h3>Methods</h3><div>In this study, Kolmogorov-Arnold networks (KANs) were used for the binary classification of QCT imaging data to assess respiratory conditions associated with cement dust exposure. The dataset comprised QCT images from 609 individuals, including 311 subjects exposed to cement dust and 298 healthy controls. We derived 141 QCT-based variables and employed KANs with two hidden layers of 15 and 8 neurons. The network parameters, including grid intervals, polynomial order, learning rate, and penalty strengths, were carefully fine-tuned. The performance of the model was assessed through various metrics, including accuracy, precision, recall, F1 score, specificity, and the Matthews Correlation Coefficient (MCC). A five-fold cross-validation was employed to enhance the robustness of the evaluation. SHAP analysis was applied to interpret the sensitive QCT features.</div></div><div><h3>Results</h3><div>The KAN model demonstrated consistently high performance across all metrics, with an average accuracy of 98.03 %, precision of 97.35 %, recall of 98.70 %, F1 score of 98.01 %, and specificity of 97.40 %. The MCC value further confirmed the robustness of the model in managing imbalanced datasets. The comparative analysis demonstrated that the KAN model outperformed traditional methods and other deep learning approaches, such as TabPFN, ANN, FT-Transformer, VGG19, MobileNets, ResNet101, XGBoost, SVM, random forest, and decision tree. SHAP analysis highlighted structural and functional lung features, such as airway geometry, wall thickness, and lung volume, as key predictors.</div></div><div><h3>Conclusion</h3><div>KANs significantly improved the classification of QCT imaging data, enhancing early detection of cement dust-induced respiratory conditions. SHAP analysis supported model interpretability, enhancing its potential for clinical translation in occupational health assessments.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103166"},"PeriodicalIF":6.1,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144185317","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":"ICD code mapping model based on clinical text tree structure","authors":"Jingjin Xue,&nbsp;Pengli Lu","doi":"10.1016/j.artmed.2025.103163","DOIUrl":"10.1016/j.artmed.2025.103163","url":null,"abstract":"<div><div>With the rapid development and progress of big data and artificial intelligence technology, the ICD coding problem of electronic medical records has been effectively solved. The deep learning method, which replaces the manual coding method, has improved the quality and efficiency of coding. However, it also faces some challenges, such as poor and fuzzy semantic representation of clinical record text and failure to consider the structural characteristics of clinical records. To address these problems, our study proposed an ICD Coding model (<strong>TR</strong>ansformer and <strong>TR</strong>ee-lstm for <strong>I</strong>CD <strong>C</strong>oding, <strong>TRIC</strong>), which enables adequate automatic ICD encoding of unstructured clinical records. In this model, the structure and features of clinical records are extracted by the constituency tree model and the transformer based model respectively, and the Tree-lstm model is used to enrich the features. Then bioBERT pre-training model is used to highlight the role of key ICD coding and improve its matching performance. Finally, it is classified by a fully connected neural network classifier to realize the many-to-many mapping between clinical records and ICD codes. On the widely used MIMIC-III full data set and sample data set, the TRIC model is compared with 12 benchmark models. The best results of 0.586, 0.109, 0.989, 0.937 and 0.758 were obtained for MiF, MaF, MiAUC, MaAUC and P@8, respectively, which verified that the TRIC model can effectively improve the quality of ICD automatic coding.</div></div>","PeriodicalId":55458,"journal":{"name":"Artificial Intelligence in Medicine","volume":"167 ","pages":"Article 103163"},"PeriodicalIF":6.1,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169826","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}