International Journal of Medical Informatics最新文献

{"title":"A web-based tool utilizing machine learning algorithms for predicting illicit drug use in emergency departments","authors":"Tsung-Chien Lu ,&nbsp;Chih-Chuan Lin ,&nbsp;Te-I Weng ,&nbsp;Fan-Ya Chou ,&nbsp;Cheng- Chung Fang ,&nbsp;TEDAS Research Group","doi":"10.1016/j.ijmedinf.2025.106031","DOIUrl":"10.1016/j.ijmedinf.2025.106031","url":null,"abstract":"<div><h3>Background</h3><div>Identifying illicit drug use through urine testing is time-consuming in the era of new psychoactive substances. This study aimed to develop a machine learning (ML) prediction model for early identification of illicit drug use in suspected emergency department (ED) patients.</div></div><div><h3>Methods</h3><div>Data from the Taiwan Emergency Department Drug Abuse Surveillance (TEDAS) database (2020–2023) was used. Six feature categories—demographics, triage data, referral source, symptoms, physical findings, and clinical characteristics—were included. The primary outcome was positive urine results for illicit drugs, confirmed by liquid chromatography-tandem mass spectrometry. Data were divided chronologically into training/validation and testing sets. Three supervised ML algorithms, including random forest, CatBoost, and light gradient boosting machine, were tested using K-fold cross-validation, and performance was evaluated by the area under the receiver operating characteristic curve (AUC) in the test set.</div></div><div><h3>Results</h3><div>The analysis included 13,615 urine test results from ED cases, identifying 3,185 positive cases (23.4%). A total of 9,529 cases (2020–2022) formed the training/validation cohort, and 4,086 (2023) were used for testing. Twenty features were used to construct the prediction model. The CatBoost classifier performed best, achieving an AUC of 0.846 (95% confidence interval [CI]: 0.831–0.859) in the testing cohort. A web-based tool and mobile apps were implemented to assist emergency physicians in predicting illicit drug use.</div></div><div><h3>Conclusions</h3><div>The machine learning model effectively predicts illicit drug use in ED patients and has been successfully implemented for free access. Further analysis is needed to assess post-implementation performance and its potential for use in other countries.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106031"},"PeriodicalIF":3.7,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144557446","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":"Project Management Digitalisation of the Clinical Research at the University Medical Centre: Good Practice of using REDCap as a Digitalisation Tool","authors":"Zdenko Garašević,&nbsp;Franc Strle,&nbsp;Martina Jaklič","doi":"10.1016/j.ijmedinf.2025.106028","DOIUrl":"10.1016/j.ijmedinf.2025.106028","url":null,"abstract":"<div><h3>Objective</h3><div>The digital tool REDCap (Research Electronic Data Capture) was implemented at the University Medical Centre Ljubljana (UMCL) with the goal of digitalising and streamlining research processes. This study aimed to assess the efficiency and transparency of clinical research following the implementation of REDCap.</div></div><div><h3>Methods</h3><div>The implementation of REDCap for funded research began in 2021. It comprised four key steps: (I) the initial creation of Central Research Registry, followed by additional functionalities including (II) the establishment of the Central Database for ’Pre-Contract Activities’ for clinical trials; (III) the integration of Reporting on Research Progress directly into the Central Research Registry; and (IV) the development of a semi-automated Workflow for internal agreements.</div></div><div><h3>Results</h3><div>Between 2021 and 2023, UMCL established a Central Research Registry using REDCap, transitioning from paper-based to digital data collection for over 2,500 research projects. These projects included clinical trials, national and international studies, as well as academic research. In addition to serving as a registry, the central system provided comprehensive data management, streamlined communication, and enhanced collaboration among stakeholders in clinical trial research at UMCL. The implementation of REDCap significantly reduced administrative burden and shortened the time required to finalise clinical trial agreements from 202 to 147 days. It also improved coordination, transparency, and real-time monitoring of research activities, facilitating more efficient research execution. Additionally, the digitalisation of internal agreements processes between researchers and stakeholders within UMCL improved coordination and expedited research execution timelines. Furthermore, REDCap enabled real-time monitoring of research progress, further contributing to the efficiency and transparency of research activities.</div></div><div><h3>Conclusion</h3><div>The digitalisation of research processes using REDCap improved the organisation and execution of research, leading to greater efficiency and transparency, reduced administrative workload, and enhanced collaboration. This approach contributed to higher-quality research outcomes and ultimately benefited patient care.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106028"},"PeriodicalIF":3.7,"publicationDate":"2025-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144519196","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":"Enhancing leukemia detection in medical imaging using deep transfer learning","authors":"Afeez A. Soladoye ,&nbsp;David B. Olawade ,&nbsp;Ibrahim A. Adeyanju ,&nbsp;Temitope Adereni ,&nbsp;Kazeem M. Olagunju ,&nbsp;Aanuoluwapo Clement David-Olawade","doi":"10.1016/j.ijmedinf.2025.106023","DOIUrl":"10.1016/j.ijmedinf.2025.106023","url":null,"abstract":"<div><h3>Background</h3><div>Acute Lymphoblastic Leukemia (ALL) is the most common pediatric cancer, requiring early detection to save lives and reduce the financial burden of advanced-stage treatment. While traditional diagnostic methods are time-consuming and resource-intensive, deep transfer learning offers a computationally efficient alternative for medical image classification.</div></div><div><h3>Method</h3><div>This study employed two widely recognized transfer learning algorithms, VGG-19 and EfficientNet-B3, to detect ALL using a publicly available dataset of 10,661 images from 118 patients. Data preprocessing included resizing, augmentation, and normalization. The models were trained for 100 epochs, with batch sizes of 30 for VGG-19 and 32 for EfficientNet-B3. Evaluation metrics such as accuracy, precision, recall, and F1 score were used to assess model performance. Statistical significance testing was performed using paired t-tests (p &lt; 0.05). Comparative analysis was performed with existing studies to validate the findings.</div></div><div><h3>Results</h3><div>EfficientNet-B3 significantly outperformed VGG-19, achieving an average accuracy of 96 % compared to 80 % for VGG-19 (p &lt; 0.001). EfficientNet-B3 demonstrated superior performance in handling class imbalance, with the minority class (Hem) achieving precision, recall, and F1 scores of 97 %, 89 %, and 93 %, respectively. VGG-19 struggled with the minority class, achieving lower recall (51 %) and F1 score (62 %). However, dataset limitations including single-source origin may affect generalizability.</div></div><div><h3>Conclusion</h3><div>This study highlights the effectiveness of EfficientNet-B3 as a reliable tool for early ALL detection, offering high accuracy and computational efficiency. Clinical implementation requires addressing computational constraints and integration challenges. Future research could integrate multimodal datasets to identify risk factors and further improve diagnostic accuracy.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106023"},"PeriodicalIF":3.7,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144501221","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":"Development of a machine learning-based prognostic model for survival prediction in patients with lung cancer brain metastases using multicenter clinical data","authors":"Yuyan Xie,&nbsp;Xuqin Xiang,&nbsp;Menglin Fan,&nbsp;Hongyan Li,&nbsp;Lijuan Du,&nbsp;Weitong Gao,&nbsp;Tong Chen,&nbsp;Zhihao Shi,&nbsp;Xinqi Yu,&nbsp;Fang Liu","doi":"10.1016/j.ijmedinf.2025.106025","DOIUrl":"10.1016/j.ijmedinf.2025.106025","url":null,"abstract":"<div><h3>Methods</h3><div>Accurate prognosis prediction for lung cancer brain metastasis (LCBM) patients is critical for clinical decision-making. This study integrates data from the SEER database (n = 2624) and Harbin Medical University Cancer Hospital (n = 362) to develop a machine learning-based prognostic prediction tool. Prognostic factors were selected through Cox regression analysis, and eight prediction models, including XGBoost, Random Forest, and Logistic Regression, were constructed. Performance was evaluated using AUC, learning curves, and PR curves, while the impact of lymph node metastasis was explored through propensity score matching and Kaplan-Meier survival analysis.</div></div><div><h3>Results</h3><div>Risk factors identified included age ≥60 years, T3 stage, and multiple organ metastases, while protective factors included female gender and household income ≥$100,000. The XGBoost model demonstrated superior performance, with mean AUCs of 0.957 (Model 1) and 0.550 (Model 2). The XGBoost-Surv model showed stable performance in both the training set (C-index = 0.653, AUC = 0.731) and the test set (C-index = 0.634, AUC = 0.705). Lymph node metastasis significantly affected prognosis (<em>p</em> &lt; 0.001), though differences in metastatic stages were not statistically significant (<em>p</em> = 0.935).</div></div><div><h3>Conclusion</h3><div>The XGBoost model developed from multicenter data effectively predicts survival outcomes in LCBM patients, with lymph node metastasis serving as an independent prognostic indicator. This model provides a reliable tool for personalized treatment decision-making.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106025"},"PeriodicalIF":3.7,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144519151","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":"Modeling unknowns: A vision for uncertainty-aware machine learning in healthcare","authors":"Andrea Campagner ,&nbsp;Elia Mario Biganzoli ,&nbsp;Clara Balsano ,&nbsp;Cristina Cereda ,&nbsp;Federico Cabitza","doi":"10.1016/j.ijmedinf.2025.106014","DOIUrl":"10.1016/j.ijmedinf.2025.106014","url":null,"abstract":"<div><div>The integration of machine learning (ML) into healthcare is accelerating, driven by the proliferation of biomedical data and the promise of data-driven clinical support. A key challenge in this context is managing the pervasive uncertainty inherent in medical reasoning and decision-making. Despite its recognized importance, uncertainty is often underrepresented in the design and evaluation of clinical AI systems.</div><div>Here we report an editorial overview of a special issue dedicated to uncertainty modeling in medical AI, which gathers theoretical, methodological, and practical contributions addressing this critical gap. Across these works, authors reveal that fewer than 4% of studies address uncertainty explicitly, and propose alternative design principles—such as optimizing for clinical net benefit or embedding explainability with confidence estimates. Notable contributions include the RelAI system for real-time prediction reliability, empirical findings on how uncertainty communication shapes clinical interpretation, and benchmarks for out-of-distribution detection in tabular data. Furthermore, this issue highlights the use of causal reasoning and anomaly detection to enhance system robustness and accountability.</div><div>Together, these studies argue that representing, communicating, and operationalizing uncertainty are essential not only for clinical safety but also for building trust in AI-driven care. This special issue thus repositions uncertainty from a limitation to a foundational asset in the responsible deployment of ML in healthcare.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106014"},"PeriodicalIF":4.1,"publicationDate":"2025-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144555881","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":"Comparing the accuracy of large language models and prompt engineering in diagnosing realworld cases","authors":"Guanhong Yao ,&nbsp;WuJi Zhang ,&nbsp;Yingxi Zhu ,&nbsp;Ut-kei Wong ,&nbsp;Yanfeng Zhang ,&nbsp;Cui Yang ,&nbsp;Guanghao Shen ,&nbsp;Zhanguo Li ,&nbsp;Hui Gao","doi":"10.1016/j.ijmedinf.2025.106026","DOIUrl":"10.1016/j.ijmedinf.2025.106026","url":null,"abstract":"<div><h3>Importance</h3><div>Large language models (LLMs) hold potential in clinical decision-making, especially for complex and rare disease diagnoses. However, real-world applications require further evaluation for accuracy and utility.</div></div><div><h3>Objective</h3><div>To evaluate the diagnostic performance of four LLMs (GPT-4o mini, GPT-4o, ERNIE, and Llama-3) using real-world inpatient medical records and assess the impact of different prompt engineering methods.</div></div><div><h3>Method</h3><div>This single-center, retrospective study was conducted at Peking University International Hospital. It involved 1,122 medical records categorized into common rheumatic autoimmune diseases, rare rheumatic autoimmune diseases, and non-rheumatic diseases. Four LLMs were evaluated using two prompt engineering methods: few-shot and chain-of-thought prompting. Diagnostic accuracy (hit1) was defined as the inclusion of the first final diagnosis from the medical record in the model’s top prediction.</div></div><div><h3>Results</h3><div>Hit1 of four LLMs were as follows: GPT-4omini (81.8 %), GPT-4o (82.4 %), ERNIE (82.9 %) and Llama-3 (82.7 %). Few-shot prompting significantly improved GPT-4o’s hit1 (85.9 %) compared to its base model (p = 0.02), outperforming other models (all p &lt; 0.05). Chain-of-thought prompting showed no significant improvement. Hit1 for both common and rare rheumatic diseases was consistently higher than that for non-rheumatic disease. Few-shot prompting increased costs per correct diagnosis for GPT-4o by approximately ¥4.54.</div></div><div><h3>Conclusions</h3><div>LLMs, including GPT-4o, demonstrate promising diagnostic accuracy on real medical records. Few-shot prompting enhances performance but at higher costs, underscoring the need for accuracy improvements and cost management. These findings inform LLM development in Chinese medical contexts and highlight the necessity for further multi-center validation.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106026"},"PeriodicalIF":3.7,"publicationDate":"2025-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144549811","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":"Artificial intelligence for solving pediatric clinical cases: A Retrieval-Augmented approach utilizing Llama3.2 and structured references","authors":"Gianluca Mondillo,&nbsp;Simone Colosimo,&nbsp;Alessandra Perrotta,&nbsp;Vittoria Frattolillo,&nbsp;Mariapia Masino,&nbsp;Marco Martino,&nbsp;Emanuele Miraglia del Giudice,&nbsp;Pierluigi Marzuillo","doi":"10.1016/j.ijmedinf.2025.106027","DOIUrl":"10.1016/j.ijmedinf.2025.106027","url":null,"abstract":"<div><h3>Background</h3><div>The “hallucinations” of Large Language Models (LLMs) raise concerns about their accuracy in pediatrics. This study aimed to evaluate whether integrating information from the Nelson Textbook of Pediatrics through a Retrieval-Augmented Generation (RAG) system could enhance the performance of Llama3.2 in addressing complex pediatric clinical cases.</div></div><div><h3>Methods</h3><div>We assessed the RAG system performance using 1,713 multiple-choice pediatric clinical questions from the MedQA dataset (n = 1,572) and Archives of Disease in Childhood–Education and Practice (n = 141). Each question was presented to Llama3.2 both in standalone mode and with RAG integration. The percentage of correct answers between models was compared using the chi-square test. p &lt; 0.05 was considered statistically significant.</div></div><div><h3>Results</h3><div>The RAG-integrated system significantly outperformed standalone Llama3.2, achieving an overall accuracy of 67.78 % (1,161/1,713) compared to 46.18 % (791/1,713) for Llama3.2 alone (p = 1.5e-112). The improvement was consistent across all pediatric subspecialties.</div></div><div><h3>Conclusions</h3><div>Incorporating RAG systems into clinical decision-making can enhance reliability and safety.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106027"},"PeriodicalIF":3.7,"publicationDate":"2025-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144365659","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":"Improving personalized healthcare with automated longitudinal EHR analysis","authors":"Gautam Pal","doi":"10.1016/j.ijmedinf.2025.106010","DOIUrl":"10.1016/j.ijmedinf.2025.106010","url":null,"abstract":"<div><div><em>Background:</em> Traditional Electronic Health Record (EHR) data analysis at King's College Hospital relies on extensive manual effort, from data extraction to reporting, limiting efficiency and scalability. This study presents an automated framework for longitudinal EHR data analysis to enhance personalized healthcare insights.</div><div><em>Methods:</em> Central to the framework is the integration of Markov Chains with Survival Analysis (SA) and Latent Growth Modeling, enhancing the modeling of patient trajectories and capturing variances in growth patterns over time. Expectation-Maximization with Gaussian Mixture Models, extended with Latent Class Analysis, identifies clinically meaningful patient subgroups for tailored interventions. The framework addresses data uncertainty, enabling precise event forecasts and trajectory predictions. The system employs Apache NiFi for data ingestion, Elasticsearch for indexing, and Splunk and Kibana for real-time visualization and reporting. Natural Language Processing (NLP) techniques extract structured insights from unstructured clinical notes, enriching datasets with context. The automation significantly reduces manual processing while ensuring data integrity and enhancing predictive capabilities.</div><div><em>Main findings:</em> Implementation demonstrated a 15% increase in detecting major depression cases, an 18% improvement in predicting patient decisions, a 25% reduction in growth trajectory prediction variance, and a 10% increase in event prediction accuracy. The framework enhances data-driven decision-making, supporting personalized healthcare interventions through real-time insights.</div><div><em>Conclusions:</em> This automated framework integrates predictive modeling, NLP techniques, and real-time data processing, improving the efficiency and accuracy of longitudinal EHR analysis. Providing robust, actionable insights enables personalized healthcare delivery, enhances clinical decision-making, and optimizes patient outcomes.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106010"},"PeriodicalIF":3.7,"publicationDate":"2025-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144480611","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":"IBDAIM:Artificial intelligence for analyzing intestinal biopsies pathological images for assisted integrated diagnostic of inflammatory bowel disease","authors":"Chengfei Cai ,&nbsp;Qianyun Shi ,&nbsp;Mingxin Liu ,&nbsp;Jun Li ,&nbsp;Yangshu Zhou ,&nbsp;Andi Xu ,&nbsp;Dan Zhang ,&nbsp;Yiping Jiao ,&nbsp;Yao Liu ,&nbsp;Xiaobin Cui ,&nbsp;Jun Chen ,&nbsp;Jun Xu ,&nbsp;Qi Sun","doi":"10.1016/j.ijmedinf.2025.106024","DOIUrl":"10.1016/j.ijmedinf.2025.106024","url":null,"abstract":"<div><h3>Background</h3><div>Inflammatory bowel disease (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC), is challenging to diagnose accurately from pathological images due to its complex histological features. This study aims to develop an artificial intelligence (AI) model, IBDAIM, to assist pathologists in quickly and accurately diagnosing IBD by analyzing whole-slide images (WSIs) of intestinal biopsies.</div></div><div><h3>Methods</h3><div>This retrospective cohort study used data from two institutions, Nanjing Drum Tower Hospital (NDTH) and Zhujiang Hospital (ZJH). The NDTH dataset was randomly divided into a model development set and an internal test set, while the ZJH dataset served as an external validation set. We developed a weakly supervised deep learning model, IBDAIM, that uses WSI-level diagnostic labels without detailed annotation. The model integrates features from patch-level predictions using Patch Likelihood Histogram (PLH) and Bag of Words (BoW) to build WSI-level representations. Performance was evaluated using area under the receiver operating characteristic curve (AUROC), accuracy (ACC), sensitivity, and specificity. Probability plots and heatmaps were generated to analyze and visualize the diagnostic labels and organizational results of WSIs. Additionally, the model was applied to assist pathologists in diagnosis, and the improvement in diagnostic performance was assessed.</div></div><div><h3>Results</h3><div>In the normal intestinal mucosa vs. IBD task, the internal test cohort achieved an AUROC of 0.998 (95% CI 0.995–1.000) and ACC of 0.982, while the external test cohorts achieved an AUROC of 0.967 (95% CI 0.939–0.995) and ACC of 0.934. For the CD vs. UC task, the internal test cohort achieved an AUROC of 0.972 (95% CI 0.942–1.000) and ACC of 0.901, and the external test cohorts achieved an AUROC of 0.952 (95% CI 0.923–0.982) and ACC of 0.949. The model’s performance exceeded that of five pathologists, and AI assistance significantly improved diagnostic accuracy across all pathologists.</div></div><div><h3>Conclusion</h3><div>The IBDAIM model demonstrates high performance in diagnosing IBD biopsy pathological images and can effectively assist pathologists in identifying normal intestinal mucosa, CD, and UC tissues. This AI tool enhances diagnostic efficiency and accuracy, supporting better clinical decision-making and patient outcomes.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106024"},"PeriodicalIF":3.7,"publicationDate":"2025-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489523","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":"Enhancing rare disease detection with deep phenotyping from EHR narratives: evaluation on Jeune syndrome","authors":"Carole Faviez ,&nbsp;Xiaomeng Wang ,&nbsp;Marc Vincent ,&nbsp;Nicolas Garcelon ,&nbsp;Sophie Saunier ,&nbsp;Valérie Cormier-Daire ,&nbsp;Xiaoyi Chen ,&nbsp;Anita Burgun","doi":"10.1016/j.ijmedinf.2025.106021","DOIUrl":"10.1016/j.ijmedinf.2025.106021","url":null,"abstract":"<div><h3>Background</h3><div>Patients with rare diseases frequently experience misdiagnoses and long diagnostic delays. Accelerating their diagnosis is essential to ensure timely access to appropriate care. Given the increasing availability of EHRs, combining artificial intelligence and deep phenotyping from large-scale clinical databases offers a promising approach to identify undiagnosed patients. This study assesses the impact of improved phenotype extraction on a screening algorithm for Jeune syndrome, a rare ciliopathy characterized by skeletal abnormalities.</div></div><div><h3>Methods</h3><div>Phenotypes from Jeune syndrome patients and controls were automatically extracted from patient unstructured EHRs relying on two thesauri separately: the standard UMLS Metathesaurus and the UMLS+, an enhanced version incorporating additional terms identified through deep learning. The machine learning pipeline that we designed for classifying patients with renal ciliopathy was adapted for Jeune syndrome detection. The model was trained and tested on both the datasets created using the two phenotyping strategies.</div></div><div><h3>Results</h3><div>Using UMLS+ strongly improved the classification of patients with Jeune syndrome, increasing the sensitivity from 49 % to 95 % while maintaining a 90 % specificity. The review of a subset of misclassified controls showed that most of them (69 %) had other genetic skeletal disorders, indicating that the model also captured patients who would benefit from referral to a bone disease geneticist.</div></div><div><h3>Conclusion</h3><div>AI-based screening combined with high-quality deep phenotyping can help reduce diagnostic delay in rare diseases. The completeness and accuracy of phenotyping from EHRs have a strong impact on screening performances.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"203 ","pages":"Article 106021"},"PeriodicalIF":3.7,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144365658","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}