Journal of Medical Systems最新文献

{"title":"Development and Internal Validation of a Vectorcardiography-Augmented Model for 12-Month Major Adverse Cardiovascular Events in Chronic Heart Failure.","authors":"Kaiyuan Cen, Zhuoqiao He, Hong Chen, Yi Tan, Lixia Lin, Xiaojuan Xu","doi":"10.1007/s10916-026-02432-y","DOIUrl":"10.1007/s10916-026-02432-y","url":null,"abstract":"<p><p>To develop and internally validate a vectorcardiography (VCG)-augmented model for estimating 12-month major adverse cardiovascular events (MACE) in hospitalized patients with chronic heart failure (CHF). We conducted a single-centre retrospective cohort study of adults hospitalized with CHF between 31 May 2023 and 31 May 2024, with data lock on 31 May 2025. The prespecified endpoint was any MACE within 12 months after the index hospitalization. ECG-to-VCG transformation was performed using the Kors method. The prespecified primary six-predictor model included LVEDD, NYHA class, frontal, horizontal, and sagittal QRS-T angles, and the QRS-loop reversal/U-turn sign. BNP and LVEF were evaluated in full-model, comparator-model, and incremental-value analyses. Because the prediction target was fixed 12-month risk rather than time-to-event hazard, multivariable logistic regression was used as the primary modelling approach. Internal validation was performed using 1000 bootstrap resamples, with apparent, optimism-corrected, calibration, and shrinkage-adjusted performance reported. Of 201 screened, 160 were included; MACE occurred in 68/160 (42.5%). The six-predictor model showed an apparent AUC of 0.946 (95% CI 0.914-0.978). Bootstrap internal validation yielded an AUC optimism estimate of 0.012 and an optimism-corrected AUC of 0.934. The apparent Brier score was 0.091, the apparent calibration slope was 1.000, and the apparent calibration intercept was 0.000. After bootstrap correction, the Brier score was 0.106, calibration-in-the-large was - 0.009, calibration intercept was - 0.010, calibration slope was 0.852, and the applied uniform shrinkage factor was 0.852. In formal incremental analyses, adding VCG features to a conventional base model improved AUC from 0.890 to 0.955 (ΔAUC 0.065; DeLong P = 0.001), with a lower Brier score and favourable discrimination/reclassification indices. A VCG-augmented model incorporating LVEDD, NYHA class, spatial QRS-T angles, and the QRS-loop reversal/U-turn sign showed promising internally validated performance for estimating 12-month MACE risk in hospitalized patients with CHF. External multicentre validation is required before clinical implementation.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148352341","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Calibration of Self-Reported Confidence and Accuracy of Large Language Models in Medical Question Answering.","authors":"Sebastian Daniel Boie, Florian Reis, Nicolas Frey, Elias Grünewald, Felix Balzer","doi":"10.1007/s10916-026-02430-0","DOIUrl":"10.1007/s10916-026-02430-0","url":null,"abstract":"<p><strong>Objective: </strong>To evaluate the alignment between self-reported confidence of large language models (LLM) and their accuracy in answering medical multiple-choice questions.</p><p><strong>Materials and methods: </strong>We prompted six LLMs (GPT-5, GPT-5-mini, GPT-5-nano, GPT-4o, Claude Sonnet 4.5, and Gemini 2.5) to answer MedMCQA items and report confidence scores. Based on 12,000 LLM responses, we calculated Expected Calibration Errors (ECE) by averaging absolute differences between observed accuracy and predicted confidence.</p><p><strong>Results: </strong>Mean ECE differed by model (Claude Sonnet 4.5 best: 0.06; Gpt-4o worst: 0.127) and varied across specialties (\"Skin\" best: 0.041; \"Social & Preventive Medicine\" worst: 0.141). Accuracy of examined LLMs showed analogous variation between specialties.</p><p><strong>Discussion: </strong>Our results demonstrate that high accuracy does not guarantee reliable uncertainty estimation. We identified substantial heterogeneity across medical specialties, where pooled metrics masked a threefold ECE increase between best- and worst-performing domains.</p><p><strong>Conclusion: </strong>We recommend incorporating calibration reporting into LLM evaluations, as larger models exhibit improved \"self-knowledge\", but uneven overconfidence persists.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13309378/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148338971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development and Validation of an Automated Acute Kidney Injury E-Alert System Integrated with Clinical Decision Support for Hospitalized Patients.","authors":"Chien-Hao Su, Chien-Ning Hsu","doi":"10.1007/s10916-026-02428-8","DOIUrl":"10.1007/s10916-026-02428-8","url":null,"abstract":"<p><p>Automated electronic alerts integrated into inpatient electronic health records (EHRs) may improve recognition of moderate-to-severe acute kidney injury (AKI) and support timely medication review. An AKI e-alert system integrated with medication-focused clinical decision support (CDS) was implemented to identify adults with KDIGO stage 2-3 hospital-acquired AKI in near real time while minimizing alert burden. The alert logic used KDIGO serum creatinine (SCr) criteria, defined the operational real-time inpatient baseline as the lowest SCr value within the preceding seven days, and applied suppression rules for dialysis within seven days or imminent transfer/discharge. Alerts were generated through scheduled near-real-time processing four times daily and were accompanied by secure in-hospital medication guidance; the CDS did not automatically place orders. System performance was evaluated at a 2,768-bed tertiary medical center in Taiwan using EHR data from March 2018 to May 2023 and validated against a retrospective computerized reference algorithm designed to replicate the deployed logic. The system generated 3,946 stage 2-3 AKI alerts, achieving 90.94% sensitivity and 99.65% accuracy compared with the reference algorithm. Seasonal alert rates were highest in winter (4.48%) and lowest in summer (3.17%), supporting the temporal face validity of the detection approach. In a hospital-wide physician survey (n = 78), 63.0% agreed that the medication guidance accompanying alerts was clinically helpful. An EHR-integrated e-alert targeting advanced AKI can achieve high real-world performance, and linking alerts to targeted medication CDS may facilitate nephrotoxin stewardship and medication safety while limiting alert fatigue.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13309393/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148338929","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Throughput Benchmarking and Throughput Variance Analysis to Evaluate the Efficiency of an Outpatient Endoscopy Unit.","authors":"Xing Li, Eli Patt, Madeline Koehn, Michael Hu, Ana Cecilia Zenteno Langle, Hrudya Viswanadh, Erin Stewart, Duncan J Flynn, Amandeep Singh","doi":"10.1007/s10916-026-02427-9","DOIUrl":"10.1007/s10916-026-02427-9","url":null,"abstract":"<p><p>Rising demand for colorectal cancer screening has increased colonoscopy volume, producing backlogs and prolonged wait times. Commonly used metrics such as room utilization provide limited insight into actionable drivers of efficiency; we applied a novel throughput benchmarking framework to quantify performance and identify operational sources of inefficiency. We conducted a pilot study in a four-room outpatient endoscopy unit from July-December 2022. Using routinely recorded timestamps for procedural stages, we derived a realistic daily target throughput (RDTT). Monthly realistic target throughput (RMTT) was calculated from RDTT. Actual monthly throughput (AMT) was benchmarked against RMTT to generate a throughput efficiency ratio (TER = AMT/RMTT) and throughput variance (TV = RMTT - AMT), which was decomposed into seven predefined operational factors. AMT ranged from 567 to 717 procedures per month, whereas RMTT ranged from 1,165 to 1,332, yielding a mean TER of 53%. The largest contributors to variance were rooms closed due to lack of endoscopist assignment (45%), workflow delays (21%), and patient no-shows or late cancellations (11%). In this pilot, we demonstrate that using commonly captured operational metrics, throughput benchmarking provides a practical framework to quantify efficiency and identify specific, actionable drivers of underperformance in endoscopy operations.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148308480","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MTA-Swin: A Multi-Token Attention Swin Transformer for Brain Tumor Classification with Leakage-Free MRI Benchmarking.","authors":"Dong Lu, Yu Zhang, Divya Chaudhary","doi":"10.1007/s10916-026-02418-w","DOIUrl":"10.1007/s10916-026-02418-w","url":null,"abstract":"<p><p>Brain tumors represent a major global health challenge, and accurate classification of brain tumors is essential for effective diagnosis and treatment. Magnetic resonance imaging (MRI) is the most commonly used and reliable modality in early brain tumor detection, and numerous studies have leveraged MRI datasets to train deep learning models for classification. However, many widely adopted brain tumor MRI datasets suffer from duplicate-induced data leakage, which can lead to artificially inflated performance metrics and unreliable model evaluation. In this study, we systematically analyze this issue and develop an automated data cleaning pipeline capable of identifying and removing duplicate scans. By applying this pipeline to a widely used public dataset, we obtain a leakage-free benchmark dataset containing 3,522 unique MRI scans, which is used for all comparative experiments. Additionally, we propose MTA-Swin, an enhanced Swin Transformer that incorporates Multi-Token Attention by re-designing the attention computation within Swin blocks. The design refines attention logits to enrich local context and enables explicit cross-head information exchange in deeper layers, while preserving Swin's hierarchical stages and windowing scheme. MTA-Swin is first pre-trained on ImageNet-1K, and then fine-tuned on our leakage-free dataset. Experimental results show that MTA-Swin reaches an overall accuracy of 98.57% over three random seeds, outperforming thirteen representative baselines. Additional stratified cross-validation and Grad-CAM analyses further support the robustness and interpretability of the proposed model. These results indicate that MTA-Swin can serve as a practical computer-aided diagnostic support model for brain tumor MRI classification.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148295465","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Beyond Validation: Operationalising Post-Deployment Surveillance of AI Medical Devices in Clinical Practice.","authors":"Jesus A Perdomo Lampignano, Aditya Kale, Shamie Kumar, Dhruv Shah, Bhargava Reddy, David J Lowe","doi":"10.1007/s10916-026-02426-w","DOIUrl":"10.1007/s10916-026-02426-w","url":null,"abstract":"<p><p>Artificial intelligence medical devices are increasingly deployed in clinical practice, yet practical approaches to post-deployment monitoring remain poorly defined. We present a structured, decision-oriented approach to monitoring within healthcare institutions, grounded in our own deployment experience. By framing surveillance as a set of interdependent decisions, this model supports effective performance assessment and governance-linked corrective action, enabling safer and more accountable integration of AI into routine clinical care.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148277722","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Effect of a Mobile App on Medication Errors During Simulated Pediatric Cardiopulmonary Resuscitation: PEDIDOSIS App vs. Conventional Methods. A Multicenter Simulation-based Randomized Controlled Trial.","authors":"Ana Martínez-Serrano, Héctor Ruiz-Rojo, Margarita Ruano Encinar, Paula García-Sánchez, Carlos Durantez-Fernández, Teresa de Rojas","doi":"10.1007/s10916-026-02425-x","DOIUrl":"10.1007/s10916-026-02425-x","url":null,"abstract":"<p><p>Despite its low incidence, pediatric cardiopulmonary arrest remains associated with high mortality and morbidity. The adequate preparation and administration of drugs can impact favourably on the outcomes of cardiopulmonary resuscitation. Nursing staff rely on cognitive aids such as paper support, calculators, or spreadsheets to ensure correct drug administration. Numerous strategies have been proposed to prevent medication errors, including the emergence of mHealth. This study evaluates the efficacy and usefulness of the PEDIDOSIS smartphone application (app) for nursing staff in a simulated pediatric emergency setting. A multicentre high-fidelity simulation-based randomized controlled trial was conducted in a pediatric emergency setting with nursing professionals from different hospitals in Spain. Randomisation groups: PEDIDOSIS app (La Paz Hospital, Spain) vs. conventional methods (such as printed or electronic medication tables, smartphones, calculators or no cognitive aid). To avoid bias due to differences in pediatric emergency experience, each group was randomly assigned the same number of participants with experience (> 5 years) and without experience (≤ 5 years). Epinephrine, bicarbonate, midazolam, and norepinephrine on two ocassions were administered in a simulation scenario. Number and magnitude of medication errors, time taken to drug administration were evaluated as measured with video recording; stress, satisfaction, and usability as measured using standardized survey tools were evaluated. One-hundred and five nurses participated. The incidence of wrong dose administration by any deviation from the correct dose was 51.5% for the control group vs. 26.4% for the PEDIDOSIS app group (p < 0.001). The OR for a wrong dose administration > 10% deviation with PEDIDOSIS app vs. conventional methods was 0.24 [0.14-0.4]), with a risk difference of 20.7%. The proportion of participants committing several medication errors (ME) was substantially lower with the app (15.1%) than with conventional methods (57.7%). Overall satisfaction regarding drug preparation was higher and stress lower in the app group. The System Usability Scale (SUS) score obtained for PEDIDOSIS app was 78.9 out of 100.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13282230/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148277649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Building a Clinically Relevant and Technically Robust Synthetic Histopathology Dataset for Breast and Gastric Cancer.","authors":"So Hyeon Lee, Young Seop Lee, Young Jae Kim, Jisup Kim, Kwang Gi Kim","doi":"10.1007/s10916-026-02421-1","DOIUrl":"10.1007/s10916-026-02421-1","url":null,"abstract":"<p><p>Data scarcity, inter-institutional stain variability, and privacy constraints are major challenges impeding the development of generalizable artificial intelligence models in digital pathology. Although recent generative adversarial network (GAN)-based synthesis approaches show promise, they struggle to preserve fine-grained nuclear morphology or maintain class-specific histological diversity. Moreover, existing diffusion-based studies have not sufficiently addressed class-conditional synthesis across heterogeneous, multi-institutional pathology data. We introduce the Efficient Pathology Diffusion Pipeline (EPDP), a class-conditional diffusion framework that enables the generation of clinically relevant, subtype-specific synthetic histopathology images for training and validating diagnostic AI models. By providing high-fidelity, subtype-specific synthetic datasets, EPDP lowers the barriers to developing robust AI diagnostic tools and supports the establishment of standardized evaluation frameworks in digital pathology. To achieve this, EPDP integrates a customized denoising U-Net that employs nuclear details, learnable class embeddings, and CycleGAN-based stain normalization with reference-guided alignment. This design explicitly targets the preservation of fine-grained nuclear morphology, class-specific histological diversity, and stain-invariant visual consistency across institutions. Multi-institutional hematoxylin and eosin whole-slide images are curated via a two-round pathology review to build subtype-labeled training and evaluation sets. Image fidelity is assessed using FID, 1-LPIPS, and SSIM, and compared against state-of-the-art models including PathDiff, PathLDM, and DiffInfinite. Diagnostic utility was evaluated using cross-domain classification (EfficientNetV2-L), and perceptual realism was assessed using a visual Turing test (VTT). The real-synthetic FID was lower than that of real-real FID by 11.0% (breast) and 8.6% (gastric). The subtype 1-LPIPS gaps were ≤ 12%, and the SSIM gaps were ≤ 0.3%. Classifiers trained only on synthetic patches matched real-trained baselines within an ~ 1% F1 score on real-image validation. Pathologists performed at near-chance levels in the VTT, with accuracies ranging from 50% to 56%.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148271903","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial Intelligence in Self-Management of Gestational Diabetes Mellitus: A Systematic Review.","authors":"Xiaoli Guo, Keying Sun, Rou Zhang, Xiaoyu Yang, Jing Zhao, Hongxia Cui, Fangyuan He","doi":"10.1007/s10916-026-02419-9","DOIUrl":"10.1007/s10916-026-02419-9","url":null,"abstract":"&lt;p&gt;&lt;p&gt;The prevalence of gestational diabetes mellitus (GDM) continues to rise, necessitating reliable and effective self-management strategies to improve maternal and neonatal outcomes. However, current self-management models face challenges, including insufficient data monitoring and analysis, delayed modifications to treatment protocols, and excessive reliance on manual processes. With the expanding application of artificial intelligence (AI) in healthcare, its potential value in the self-management of GDM has attracted increasing attention. This systematic review aimed to synthesize the evidence on the application of AI technologies in the self-management of patients with GDM. This systematic review was conducted in April 2025 and included comprehensive literature searches across PubMed, Embase, The Cochrane Library, Scopus, Web of Science, CINAHL, CBM, CNKI, VIP, and Wanfang databases. The search strategy combined Medical Subject Headings and free-text terms related to GDM, AI, machine learning, and self-management. Quantitative studies that explored the application of AI in the self-management of patients with GDM were included, including randomized controlled trials and cohort studies. Two researchers independently performed study selection and data extraction, followed by quality assessment using risk-of-bias instruments appropriate for each study design. Data were synthesized using a narrative approach combined with thematic synthesis. The initial search yielded 18,973 records. After stepwise screening, 10 studies were included. A total of 645 patients with GDM completed AI-assisted interventions (from 661 initially enrolled), along with 864 control participants (from 877 enrolled). A variety of AI technologies were employed, including expert systems, machine learning, and natural language processing. Their primary functions included abnormality detection and alert triggering, personalized treatment plan generation and adjustment, and data integration and management. The studies reported multiple outcomes. Regarding health outcomes, six studies reported that AI interventions were associated with improved glycemic control, although heterogeneity was observed in delivery outcomes and insulin utilization rates. In terms of adherence, AI interventions tended to increase the frequency of blood glucose monitoring and data upload rates. Regarding system usability, limited data suggested that the accuracy of dietary recommendations and detection of blood glucose abnormalities was satisfactory, whereas the adoption rate of insulin treatment adjustment recommendations was relatively low. User satisfaction was generally high. Facilitators for implementation included technological advantages, user experience, and external support, whereas barriers included data integration and quality issues, technical and hardware or software limitations, patient acceptance, and difficulties in clinical integration. Preliminary evidence suggests that AI may contribu","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148226088","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MEL-IA: An Interoperable AI System for Multimodal Skin Lesion Classification in Hospital Settings.","authors":"Pablo Candela Córcoles, Alberto De Ramón Fernández, Marcelo Saval Calvo, Jose María Salinas Serrano, Diego Guijarro Peral, Daniel Ruiz Fernández","doi":"10.1007/s10916-026-02424-y","DOIUrl":"10.1007/s10916-026-02424-y","url":null,"abstract":"<p><p>Skin cancer represents a growing healthcare challenge, motivating the development of AI sytems that can support lesion assessment while integrating efficiently into clinical workflows. This study presents MEL-IA (MobilE skin Lesion dIAgnosis), an interoperable AI system designed for automated skin lesion classification and full integration with hospital information systems.</p><p><strong>Methods: </strong>The system combines dermatoscopic images and structured clinical metadata using an EfficientNet-B4 - based multimodal model trained on the ISIC 2019 BCN_20000 and MSK subsets. Internal robustness was assessed through stratified 5-fold cross-validation, and external generalization was evaluated on the independent HAM10000 dataset. Interoperability and operational performance were validated through deployment in a real hospital environment using HL7, DICOM, PACS, and HIS/RIS systems.</p><p><strong>Results: </strong>We evaluated internal robustness using stratified five‑fold cross‑validation, with global accuracy of 0.86, macro F1-score of 0.85, and AUC values above 0.97. The external evaluation showed that the model generalizes well under different acquisition conditions, achieving an accuracy of 0.84 and balanced accuracy of 0.65, with sensitivities of 0.60 for melanoma 0.72 for basal cell carcinoma. The multimodal model outperformed the image model baseline across all classes. Technical deployment confirmed full interoperability, > 99% successful study integration, and near real time processing.</p><p><strong>Conclusions: </strong>MEL-IA provides multimodal skin lesion classification capabilities and integrates smoothly into hospital infrastructures based on standards. The results demonstrate technical feasibility, interoperability, and operational viability for deployment within dermatology oriented clinical workflows. Further clinical validation studies are warranted to assess the system's impact on diagnostic decision-making and patient outcomes.</p>","PeriodicalId":16338,"journal":{"name":"Journal of Medical Systems","volume":"50 1","pages":""},"PeriodicalIF":8.8,"publicationDate":"2026-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13260036/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148226124","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}