BMC Medical Informatics and Decision Making最新文献

{"title":"Combining clinical and left atrial electromechanical remodelling data: potential to improve atrial fibrillation ablation outcome prediction.","authors":"Neil Bodagh, Iain Sim, Mahta Haghighat Ghahfarokhi, Kyaw Soe Tun, Irum Kotadia, Magda Klis, Vinush Vigneswaran, Jose Alonso Solis Lemus, John Whitaker, Ali Gharaviri, Pier-Giorgio Masci, Amedeo Chiribiri, Steven Niederer, Mark O'Neill, Steven E Williams","doi":"10.1186/s12911-025-03200-7","DOIUrl":"10.1186/s12911-025-03200-7","url":null,"abstract":"","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"356"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12482448/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191250","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":"SPA-IoT with MCSV-CNN: a novel IoT-enabled method for robust pre-ictal seizure prediction.","authors":"Dhanalekshmi Prasad Yedurkar, Shilpa P Metkar, Thompson Stephan, Vijay Mohan, Saurabh Agarwal","doi":"10.1186/s12911-025-03191-5","DOIUrl":"10.1186/s12911-025-03191-5","url":null,"abstract":"<p><p>This paper introduces a new approach to real-time epileptic seizure prediction using a lightweight Convolutional Neural Network (CNN) architecture and multiresolution feature extraction from electroencephalogram (EEG) recordings. Multiresolution Critical Spectral Verge CNN (MCSV-CNN), the suggested model, is best suited for use in wearable technology that is connected to the Internet of Things (IoT). The software module uses pre-ictal and inter-ictal EEG segments to forecast seizures early, and the signal acquisition module collects EEG data. Multiscale frequency analysis and spatial feature learning are combined in the MCSV-CNN architecture to capture minute signal changes that precede seizures. Both actual clinical EEG recordings and the Temple University Hospital EEG Seizure Corpus (TUH-EEG) were evaluated. Predicting has been performed using a 5-minute pre-ictal window and a 10-minute seizure occurrence prediction (SOP) horizon. The approach proposed outperformed a number of existing CNN-based seizure prediction techniques with an average prediction accuracy of 99.5%, sensitivity of 98.3%, false prediction rate of 0.045, and a high Area Under the Curve (AUC). These findings show that MCSV-CNN has the potential to be a dependable, real-time seizure prediction tool that could be used practically in wearable medical technology. The prediction accuracy and lightweight architecture of the technology point to its potential application in early clinical intervention and ongoing at-home monitoring.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"354"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12482214/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191368","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":"Automating the extraction of otology symptoms from clinic letters: a methodological study using natural language processing.","authors":"Nikhil Joshi, Kawsar Noor, Xi Bai, Marina Forbes, Talisa Ross, Liam Barrett, Richard J B Dobson, Anne G M Schilder, Nishchay Mehta, Watjana Lilaonitkul","doi":"10.1186/s12911-025-03180-8","DOIUrl":"10.1186/s12911-025-03180-8","url":null,"abstract":"<p><strong>Background: </strong>Most healthcare data is in an unstructured format that requires processing to make it usable for research. Generally, this is done manually, which is both time-consuming and poorly scalable. Natural language processing (NLP) using machine learning offers a method to automate data extraction. In this paper we describe the development of a set of NLP models to extract and contextualise otology symptoms from free text documents.</p><p><strong>Methods: </strong>A dataset of 1,148 otology clinic letters written between 2009 - 2011, from a London NHS hospital, were manually annotated and used to train a hybrid dictionary and machine learning NLP model to identify six key otological symptoms: hearing loss, impairment of balance, otalgia, otorrhoea, tinnitus and vertigo. Subsequently, a set of Bidirectional-Long-Short-Term-Memory (Bi-LSTM) models were trained to extract contextual information for each symptom, for example, defining the laterality of the ear affected.</p><p><strong>Results: </strong>There were 1,197 symptom annotations and 2,861 contextual annotations with 24% of patients presenting with hearing loss. The symptom extraction model achieved a macro F1 score of 0.73. The Bi-LSTM models achieved a mean macro F1 score of 0.69 for the contextualisation tasks.</p><p><strong>Conclusion: </strong>NLP models for symptom extraction and contextualisation were successfully created and shown to perform well on real life data. Refinement is needed to produce models that can run without manual review. Downstream applications for these models include deep semantic searching in electronic health records, cohort identification for clinical trials and facilitating research into hearing loss phenotypes. Further testing of the external validity of the developed models is required.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"353"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12482202/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191205","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":"Association rule mining of time-based patterns in diabetes-related comorbidities on imbalanced data: a pre- and post-diagnosis study.","authors":"Róbert Bata, Amr Sayed Ghanem, Eszter Vargáné Faludi, Ferenc Sztanek, Attila Csaba Nagy","doi":"10.1186/s12911-025-03206-1","DOIUrl":"10.1186/s12911-025-03206-1","url":null,"abstract":"<p><p>Type 2 diabetes mellitus (T2DM) is affecting over 529 million adults and anticipated to impact 1.3 billion by 2050. This disease often coexists with multiple comorbidities, which can complicate its management. These comorbidities not only increase morbidity and mortality but also challenge the effectiveness of interventions designed to manage diabetes and improve patient outcomes. We analysed imbalanced data of 25.065 patients deriving from the Clinical Centre of the University of Debrecen, Hungary. The aim of the study was to explore the prevalence and temporal patterns of comorbidities before and after the diagnosis of T2DM using Association Rule Mining (ARM) and network visualization. The initial five years following T2DM diagnosis mark a spike in newly emerging health conditions. Hypertension frequently occurs at an earlier stage, while pneumonia, eye-related disorders, and ischemic heart disease consistently appear throughout the progression of the disease. The ARM analysis showed that both acute and chronic kidney diseases, as well as respiratory disorders are common after T2DM diagnosis. Certain gender-specific trends, such as higher instances of heart failure and acute kidney injury in males, are also notable. The study highlights how ARM techniques reveal complex patterns in chronic disease management, suggesting potential pathways for targeted treatments.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"352"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12482723/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191245","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":"Ethical implications of using general-purpose LLMs in clinical settings: a comparative analysis of prompt engineering strategies and their impact on patient safety.","authors":"Pouyan Esmaeilzadeh","doi":"10.1186/s12911-025-03182-6","DOIUrl":"10.1186/s12911-025-03182-6","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;The rapid integration of large language models (LLMs) into healthcare raises critical ethical concerns regarding patient safety, reliability, transparency, and equitable care delivery. Despite not being trained explicitly on medical data, individuals increasingly use general-purpose LLMs to address medical questions and clinical scenarios. While prompt engineering can optimize LLM performance, its ethical implications for clinical decision-making remain underexplored. This study aimed to evaluate the ethical dimensions of prompt engineering strategies in the clinical applications of LLMs, focusing on safety, bias, transparency, and their implications for the responsible implementation of AI in healthcare.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We conducted an ethics-focused analysis of three advanced and reasoning-capable LLMs (OpenAI O3, Claude Sonnet 4, Google Gemini 2.5 Pro) across six prompt engineering strategies and five clinical scenarios of varying ethical complexity. Six expert clinicians evaluated 90 responses using domains that included diagnostic accuracy, safety assessment, communication, empathy, and ethical reasoning. We specifically analyzed safety incidents, bias patterns, and transparency of reasoning processes.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Significant ethical concerns emerged across all models and scenarios. Critical safety issues occurred in 12.2% of responses, with concentration in complex ethical scenarios (Level 5: 23.1% vs. Level 1: 2.3%, p &lt; 0.001). Meta-cognitive prompting demonstrated superior ethical reasoning (mean ethics score: 78.3 ± 9.1), while safety-first prompting reduced safety incidents by 45% compared to zero-shot approaches (8.9% vs. 16.2%). However, all models showed concerning deficits in communication empathy (mean 54% of maximum) and exhibited potential bias in complex multi-cultural scenarios. Transparency varied significantly by prompt strategy, with meta-cognitive approaches providing the clearest reasoning pathways (4.2 vs. 1.8 explicit reasoning steps), which are essential for clinical accountability. The study highlighted critical gaps in ethical decision-making transparency, with meta-cognitive approaches providing 4.2 explicit reasoning steps compared to 1.8 in zero-shot methods (p &lt; 0.001). Bias patterns disproportionately affected vulnerable populations, with systematic underestimation of treatment appropriateness in elderly patients and inadequate cultural considerations in end-of-life scenarios.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;Current clinical applications of general-purpose LLMs present substantial ethical challenges requiring urgent attention. While structured prompt engineering demonstrated measurable improvements in some domains, with meta-cognitive approaches showing 13.0% performance gains and safety-first prompting reducing critical incidents by 45%, substantial limitations persist across all strategies. Even optimized approaches achieved inadequate perf","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"342"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12481957/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191312","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":"Prediction of atrial fibrillation admissions in arrhythmia naïve patients from structured electronic health record data.","authors":"Tanmay Gokhale, Nirav R Bhatt, Matthew Starr, Suresh Mulukutla, Floyd Thoma, Murat Akcakaya, Salah Al-Zaiti, Raul G Nogueira, Samir Saba","doi":"10.1186/s12911-025-03199-x","DOIUrl":"10.1186/s12911-025-03199-x","url":null,"abstract":"<p><strong>Background: </strong>Atrial fibrillation (AF) is the most prevalent sustained arrhythmia, but its diagnosis is often elusive. In this study, we examined the role of machine learning (ML) algorithms in predicting AF in arrhythmia-naïve patients, based on structured domains of the electronic health records (EHR).</p><p><strong>Methods: </strong>Patients (N = 186,769) with no prior history of AF, who received at least 1 echocardiogram and who had a minimum of 3 months of follow-up, were included. Data from the EHR were grouped into domains (demographic; social determinants of health; past medical history, medications, electrocardiogram (EKG), and echocardiogram (Echo)) and tested incrementally for their ability to predict incident AF admission to the hospital.</p><p><strong>Results: </strong>Of the overall cohort, 4,751 (2.5%) patients were admitted for AF over a median follow-up time of 35 months. Incremental EHR domains increased the area under the receiver-operator curve (AUROC) for all ML classifiers, with Gradient Boosting achieving an AUROC of 0.85 when all domains were included, but with a poor F1 score of 14% at the maximal Youden index. Using the EKG and Echo domains alone achieved comparable performance to when all EHR domains were included. These results were externally validated.</p><p><strong>Conclusion: </strong>More domains of structured EHR improve the ability to predict incident AF admissions but structured EKG and Echo domains realize the most gain. Although ML models exhibited good discrimination, the precision is poor due to the low event rate.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"348"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12482350/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191327","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":"Decision fatigue of surrogate decision-makers: a scoping review.","authors":"Shasha Cai, Mei Zhang, Zhanghui Song, Yangmengyuan He, Mengchen Huang, Qinyan Shuai","doi":"10.1186/s12911-025-03198-y","DOIUrl":"10.1186/s12911-025-03198-y","url":null,"abstract":"","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"343"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12481925/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191201","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":"A study on factors influencing decision - making conflicts of proxies of thrombolytic patients with acute ischemic stroke: a network analysis.","authors":"Dazhen Li, Yuanli Guo, Keke Ma, Wenfeng Fan, Caixia Yang, Renke Gao, Li-Na Guo, Xiaofang Dong, Peihua Lv, Yanjin Liu","doi":"10.1186/s12911-025-03187-1","DOIUrl":"10.1186/s12911-025-03187-1","url":null,"abstract":"","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"346"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12481955/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191225","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 of a prediction model for lower limb deformity in patients with hereditary multiple exostoses based on interpretable models and nomogram.","authors":"Rongbin Lu, He Ling, Zhao Huang, Wencai Li, Junjie Liu, Yonghui Lao, Qian Liu, Xiaofei Ding","doi":"10.1186/s12911-025-03196-0","DOIUrl":"10.1186/s12911-025-03196-0","url":null,"abstract":"<p><strong>Aim: </strong>To investigate the impact of relevant factors on the occurrence of genu valgum and ankle valgus at admission in patients with hereditary multiple exostoses (HME) based on blood test results from the three months prior to admission, and to develop an interpretable machine learning model and nomogram to study the influence of these factors.</p><p><strong>Method: </strong>This retrospective study collected blood test results and relevant clinical data from 140 HME patients in the three months prior to admission. The data were divided into training and validation sets at a 7:3 ratio. Five machine learning models were compared to select the optimal model for explaining risk factors. Multivariate regression analysis was further used to screen independent predictors, and a nomogram for predicting the probability of lower limb deformities in HME was constructed using R and JD_DCPM (V6.03, Jingding Medical Technology Co., Ltd.).</p><p><strong>Result: </strong>The results showed that the Random Forest model demonstrated high stability in explaining the outcomes of genu valgum and ankle valgus in HME patients. SHAP analysis indicated that PA made a significant contribution to both outcomes. Multivariate regression analysis further identified ALB (0.037 [0.003-0.203]), GLB (0.083 [0.010-0.416]), and PA (0.025 [0.002-0.137]) as independent predictors for genu valgum. For ankle valgus, GLB (0.183 [0.053-0.571]), PA (0.162 [0.035-0.631]), and UA (7.156 [1.841-34.03]) were identified as independent predictors. The nomogram exhibited good prediction performance with moderate errors in both the training and validation groups.</p><p><strong>Conclusion: </strong>ALB, GLB, and PA are independent predictors of genu valgum in HME patients three months later, while GLB, PA, and UA are independent predictors of ankle valgus. The interpretable RF model and constructed nomogram in this study enable clinicians to assess the risk of lower limb deformities in HME patients as early as possible, benefiting more patients.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"355"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12482412/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191217","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":"Can large language models follow guidelines? A comparative study of ChatGPT-4o and DeepSeek AI in clavicle fracture management based on AAOS recommendations.","authors":"Tolga Keçeci, Bekir Karagöz","doi":"10.1186/s12911-025-03202-5","DOIUrl":"10.1186/s12911-025-03202-5","url":null,"abstract":"","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"350"},"PeriodicalIF":3.8,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12482356/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145191219","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}