BMC Medical Informatics and Decision Making最新文献

{"title":"Prediction of adverse pregnancy outcomes using machine learning techniques: evidence from analysis of electronic medical records data in Rwanda.","authors":"Muzungu Hirwa Sylvain, Emmanuel Christian Nyabyenda, Melissa Uwase, Isaac Komezusenge, Fauste Ndikumana, Innocent Ngaruye","doi":"10.1186/s12911-025-02921-z","DOIUrl":"10.1186/s12911-025-02921-z","url":null,"abstract":"<p><strong>Background: </strong>Despite substantial progress in maternal and neonatal health, Rwanda's mortality rates remain high, necessitating innovative approaches to meet health related Sustainable Development Goals (SDGs). By leveraging data collected from Electronic Medical Records, this study explores the application of machine learning models to predict adverse pregnancy outcomes, thereby improving risk assessment and enhancing care delivery.</p><p><strong>Methods: </strong>This study utilized retrospective cohort data from the electronic medical record (EMR) system of 25 hospitals in Rwanda from 2020 to 2023. The independent variables included socioeconomic status, health status, reproductive health, and pregnancy-related factors. The outcome variable was a binary composite feature that combined adverse pregnancy outcomes in both the mother and the newborn. Extensive data cleaning was performed, with missing values addressed through various strategies, including the exclusion of variables and instances, imputation techniques using K-Nearest Neighbors and Multiple Imputation by Chained Equations. Data imbalance was managed using a synthetic minority oversampling technique. Six machine learning models-Logistic Regression, Decision Trees, Support Vector Machine, Gradient Boosting, Random Forest, and Multilayer Perceptron-were trained using 10-fold cross-validation and evaluated on an unseen dataset with-70 - 30 training and evaluation splits.</p><p><strong>Results: </strong>Data from 117,069 women across 25 hospitals in Rwanda were analyzed, leading to a final dataset of 32,783 women after removing entries with significant missing values. Among these women, 5,424 (16.5%) experienced adverse pregnancy outcomes. Random Forest and Gradient Boosting Classifiers demonstrated high accuracy and precision. After hyperparameter tuning, the Random Forest model achieved an accuracy of 90.6% and an ROC-AUC score of 0.85, underscoring its effectiveness in predicting adverse outcomes. However, a recall rate of 46.5% suggests challenges in detecting all the adverse cases. Key predictors of adverse outcomes identified in this study included gestational age, number of pregnancies, antenatal care visits, maternal age, vital signs, and delivery methods.</p><p><strong>Conclusions: </strong>This study recommends enhancing EMR data quality, integrating machine learning into routine practice, and conducting further research to refine predictive models and address evolving pregnancy outcomes. In addition, this study recommends the design of AI-based interventions for high-risk pregnancies.</p><p><strong>Clinical trial number: </strong>Not applicable.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"76"},"PeriodicalIF":3.3,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11823242/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143405373","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":"Haematology dimension reduction, a large scale application to regular care haematology data.","authors":"Huibert-Jan Joosse, Chontira Chumsaeng-Reijers, Albert Huisman, Imo E Hoefer, Wouter W van Solinge, Saskia Haitjema, Bram van Es","doi":"10.1186/s12911-025-02899-8","DOIUrl":"10.1186/s12911-025-02899-8","url":null,"abstract":"<p><strong>Background: </strong>The routine diagnostic process increasingly entails the processing of high-volume and high-dimensional data that cannot be directly visualised. This processing may provide scaling issues that limit the implementation of these types of data into research as well as integrated diagnostics in routine care. Here, we investigate whether we can use existing dimension reduction techniques to provide visualisations and analyses for a complete bloodcount (CBC) while maintaining representativeness of the original data. We considered over 3 million CBC measurements encompassing over 70 parameters of cell frequency, size and complexity from the UMC Utrecht UPOD database. We evaluated PCA as an example of a linear dimension reduction techniques and UMAP, TriMap and PaCMAP as non-linear dimension reduction techniques. We assessed their technical performance using quality metrics for dimension reduction as well as biological representation by evaluating preservation of diurnal, age and sex patterns, cluster preservation and the identification of leukemia patients.</p><p><strong>Results: </strong>We found that, for clinical hematology data, PCA performs systematically better than UMAP, TriMap and PaCMAP in representing the underlying data. Biological relevance was retained for periodicity in the data. However, we also observed a decrease in predictive performance of the reduced data for both age and sex, as well as an overestimation of clusters within the reduced data. Finally, we were able to identify the diverging patterns for leukemia patients after use of dimensionality reduction methods.</p><p><strong>Conclusions: </strong>We conclude that for hematology data, the use of unsupervised dimension reduction techniques should be limited to data visualization applications, as implementing them in diagnostic pipelines may lead to decreased quality of integrated diagnostics in routine care.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"75"},"PeriodicalIF":3.3,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11823074/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143405262","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 novel method for screening malignant hematological diseases by constructing an optimal machine learning model based on blood cell parameters.","authors":"Dehua Sun, Wei Chen, Jun He, Yongjian He, Haoqin Jiang, Hong Jiang, Dandan Liu, Lu Li, Min Liu, Zhigang Mao, Chenxue Qu, Linlin Qu, Ziyong Sun, Jianbiao Wang, Wenjing Wu, Xuefeng Wang, Wei Xu, Ying Xing, Chi Zhang, Jingxian Zhang, Lei Zheng, Shihong Zhang, Bo Ye, Ming Guan","doi":"10.1186/s12911-025-02892-1","DOIUrl":"10.1186/s12911-025-02892-1","url":null,"abstract":"<p><strong>Background: </strong>Screening of malignant hematological diseases is of great importance for their diagnosis and subsequent treatment. This study constructed an optimal screening model for malignant hematological diseases based on routine blood cell parameters.</p><p><strong>Methods: </strong>The venous blood samples of 1751 patients collected from 10 tertiary hospitals in China were divided into a training set (1223 cases) and a validation set (528 cases). In addition to the clinical diagnostic information of the samples in the training set, 26 blood cell parameters including morphological parameters were selected using manual screening and filtering to construct eight machine learning models. These models were used to identify hematological malignancies among the validation set.</p><p><strong>Results: </strong>Comparison of the discrimination, calibration and clinical detection performance of the eight machine learning models revealed that the artificial neural network (ANN) model performed the optimal in identifying malignant haematological diseases in the validation set (528 cases), with an area under the receiver operating characteristic curve (AUC), accuracy, sensitivity and specificity of 0.906, 0.857, 0.832 and 0.884, respectively.</p><p><strong>Conclusion: </strong>The ANN model constructed can be used for screening of malignant hematological diseases, especially in primary hospitals that lack comprehensive diagnosis, and this ANN model will help patients to get diagnosis and treatment of malignant hematological diseases as early as possible.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"72"},"PeriodicalIF":3.3,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11816569/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143397961","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":"Classifying and fact-checking health-related information about COVID-19 on Twitter/X using machine learning and deep learning models.","authors":"Elham Sharifpoor, Maryam Okhovati, Mostafa Ghazizadeh-Ahsaee, Mina Avaz Beigi","doi":"10.1186/s12911-025-02895-y","DOIUrl":"10.1186/s12911-025-02895-y","url":null,"abstract":"<p><strong>Background: </strong>Despite recent progress in misinformation detection methods, further investigation is required to develop more robust fact-checking models with particular consideration for the unique challenges of health information sharing. This study aimed to identify the most effective approach for detecting and classifying reliable information versus misinformation health content shared on Twitter/X related to COVID-19.</p><p><strong>Methods: </strong>We have used 7 different machine learning/deep learning models. Tweets were collected, processed, labeled, and analyzed using relevant keywords and hashtags, then classified into two distinct datasets: \"Trustworthy information\" versus \"Misinformation\", through a labeling process. The cosine similarity metric was employed to address oversampling the minority of the Trustworthy information class, ensuring a more balanced representation of both classes for training and testing purposes. Finally, the performance of the various fact-checking models was analyzed and compared using accuracy, precision, recall, and F1-score ROC curve, and AUC.</p><p><strong>Results: </strong>For measures of accuracy, precision, F1 score, and recall, the average values of TextConvoNet were found to be 90.28, 90.28, 90.29, and 0.9030, respectively. ROC AUC was 0.901.\"Trustworthy information\" class achieved an accuracy of 85%, precision of 93%, recall of 86%, and F1 score of 89%. These values were higher than other models. Moreover, its performance in the misinformation category was even more impressive, with an accuracy of 94%, precision of 88%, recall of 94%, and F1 score of 91%.</p><p><strong>Conclusion: </strong>This study showed that TextConvoNet was the most effective in detecting and classifying trustworthy information V.S misinformation related to health issues that have been shared on Twitter/X.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"73"},"PeriodicalIF":3.3,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11817542/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143397971","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 novel method for assessing cycling movement status: an exploratory study integrating deep learning and signal processing technologies.","authors":"Yingchun He, Yi-Haw Jan, Fan Yang, Yunru Ma, Chun Pei","doi":"10.1186/s12911-024-02828-1","DOIUrl":"10.1186/s12911-024-02828-1","url":null,"abstract":"<p><p>This study proposes a deep learning-based motion assessment method that integrates the pose estimation algorithm (Keypoint RCNN) with signal processing techniques, demonstrating its reliability and effectiveness.The reliability and validity of this method were also verified.Twenty college students were recruited to pedal a stationary bike. Inertial sensors and a smartphone simultaneously recorded the participants' cycling movement. Keypoint RCNN(KR) algorithm was used to acquire 2D coordinates of the participants' skeletal keypoints from the recorded movement video. Spearman's rank correlation analysis, intraclass correlation coefficient (ICC), error analysis, and t-test were conducted to compare the consistency of data obtained from the two movement capture systems, including the peak frequency of acceleration, transition time point between movement statuses, and the complexity index average (CIA) of the movement status based on multiscale entropy analysis.The KR algorithm showed excellent consistency (ICC_1,3=0.988) between the two methods when estimating the peak acceleration frequency. Both peak acceleration frequencies and CIA metrics estimated by the two methods displayed a strong correlation (r > 0.70) and good agreement (ICC_2,1>0.750). Additionally, error values were relatively low (MAE = 0.001 and 0.040, MRE = 0.00% and 7.67%). Results of t-tests showed significant differences (p = 0.003 and 0.030) for various acceleration CIAs, indicating our method could distinguish different movement statuses.The KR algorithm also demonstrated excellent intra-session reliability (ICC = 0.988). Acceleration frequency analysis metrics derived from the KR method can accurately identify transitions among movement statuses. Leveraging the KR algorithm and signal processing techniques, the proposed method is designed for individualized motor function evaluation in home or community-based settings.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"71"},"PeriodicalIF":3.3,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11817045/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143397954","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 minimum data set and electronic registry for hemodialysis patients management.","authors":"Mahtab Karami, Ehsan Nabovati, Nasim Mirpanahi","doi":"10.1186/s12911-025-02914-y","DOIUrl":"10.1186/s12911-025-02914-y","url":null,"abstract":"<p><strong>Background: </strong>This study aimed to develop a minimum dataset and an electronic registry system for hemodialysis patients to evaluate hemodialysis patients' treatment procedures and outcomes, conduct related research, and design therapeutic interventions.</p><p><strong>Methods: </strong>This developmental research was performed in multiple phases, including content determination using the Delphi technique; database designing using MySQL; building a user interface using PHP; usability evaluation using the think-aloud method by 10 evaluators through a scenario consisting of 7 tasks; and finally, the system was piloted by entering the 160 patients' paper records into the system.</p><p><strong>Results: </strong>Following the CVR and CVI content validity assessment, 108 of the 118 extracted data elements (DEs) were validated. Then, using the Delphi technique, nephrologists chose 57 DEs and divided them into 4 information categories, including the patient's clinical history, hemodialysis episodes, laboratory findings, and the outcomes of hemodialysis. The three tabs that made up the user interface were the homepage, information recording, reports, and definitions. The problems with appearance and performance were discovered using the think-aloud method, and they were then resolved. Finally, users had the opportunity to identify issues, improve the system's capabilities, and express their satisfaction throughout the system's three-month test period.</p><p><strong>Conclusions: </strong>The E-hemodialysis registry was created based on knowledge gained from industrialized nations, opinions and suggestions from medical specialists, and the facilities that were accessible. Information from this system can be utilized as a starting point for evaluating the hemodialysis patients' status, identifying problems, and making sensible decisions for the best possible planning and management of end-stage renal disease.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"69"},"PeriodicalIF":3.3,"publicationDate":"2025-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11812217/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143390159","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":"Correction: Machine learning predicts pulmonary long Covid sequelae using clinical data.","authors":"Ermanno Cordelli, Paolo Soda, Sara Citter, Elia Schiavon, Christian Salvatore, Deborah Fazzini, Greta Clementi, Michaela Cellina, Andrea Cozzi, Chandra Bortolotto, Lorenzo Preda, Luisa Francini, Matteo Tortora, Isabella Castiglioni, Sergio Papa, Diego Sona, Marco Alì","doi":"10.1186/s12911-025-02918-8","DOIUrl":"10.1186/s12911-025-02918-8","url":null,"abstract":"","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"68"},"PeriodicalIF":3.3,"publicationDate":"2025-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11812242/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143390157","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":"Synthesis of the clinical utilities and issues of intraoperative imaging devices in clinical reports: a systematic review and thematic synthesis.","authors":"Hiroyuki Suzuki, Yusuke Tsuboko, Manabu Tamura, Ken Masamune, Kiotaka Iwasaki","doi":"10.1186/s12911-025-02915-x","DOIUrl":"10.1186/s12911-025-02915-x","url":null,"abstract":"<p><strong>Background: </strong>Intraoperative imaging devices (i-ID), such as intraoperative optical coherence tomography (iOCT), offer surgeons critical insights previously unobservable, enhancing surgical precision and safety. Despite their benefits, i-IDs present challenges that necessitate early identification and synthesis of clinical issues to promote safer surgical implementation. This study aims to explore the potential of Qualitative Evidence Synthesis (QES) for synthesising qualitative evidence from clinical reports regarding the clinical utility and issues associated with iOCT devices.</p><p><strong>Methods: </strong>In June 2022, we conducted a systematic literature search using PubMed, Web of Science, Embase, and the Cochrane Library for articles on iOCT for retinal surgery. Criteria included articles in English, with at least ten cases, and providing qualitative insights into iOCT's utilities and issues. We performed thematic synthesis from the identified articles using qualitative data analysis software, beginning with initial coding of the 'Results' and 'Discussion' sections to create themes reflecting iOCT's utilities and issues. The created themes were further refined through axial coding and were used to construct a model illustrating iOCT's potential influence on patient outcomes. The reliability and validity of the themes were ensured through independent coding, expert consultations, and iterative revisions to achieve consensus among reviewers.</p><p><strong>Results: </strong>The QES approach enabled systematic data extraction and synthesis, providing a comprehensive view of both the utilities and issues associated with iOCT. Our findings emphasise the significant role of iOCT in enhancing decision-making, specifically in membrane peeling tasks and in detecting preoperatively undetected conditions such as full-thickness macular holes. This study also revealed critical insights into the technical challenges associated with iOCT, including device malfunctions and procedural interruptions, which are vital for improving device safety and integration into surgical practice.</p><p><strong>Conclusion: </strong>The application of QES facilitated a thorough investigation into the clinical utilities and issues of iOCT, encouraging the application of this method in the ongoing evaluation of i-ID technologies. This initial experience with QES confirms its potential in synthesising qualitative clinical data and suggests its applicability to other i-ID modalities. This approach enhances the reliability of findings and provides a solid foundation for assessing clinical utilities and issues for policymakers and medical specialists.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"70"},"PeriodicalIF":3.3,"publicationDate":"2025-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11812201/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143390160","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":"Bayesian learning-based agent negotiation model to support doctor-patient shared decision making.","authors":"Xin Chen, Yong Liu, Fei-Ping Hong, Ping Lu, Jiang-Tao Lu, Kai-Biao Lin","doi":"10.1186/s12911-024-02839-y","DOIUrl":"10.1186/s12911-024-02839-y","url":null,"abstract":"<p><strong>Background: </strong>Agent negotiation is widely used in e-commerce negotiation, cloud service service-level agreements, and power transactions. However, few studies have adapted alternative negotiation models to negotiation processes between healthcare professionals and patients due to the fuzziness, ethics, and importance of medical decision making.</p><p><strong>Method: </strong>We propose a Bayesian learning based bilateral fuzzy constraint agent negotiation model (BLFCAN). It support mutually beneficial agreement on treatment between doctors and patients. The proposed model expresses the imprecise preferences and behaviors of doctors and patients through fuzzy constrained agents. To improve negotiation efficiency and social welfare, the Bayesian learning method is adopted in the proposed model to predict the opponent's preference.</p><p><strong>Results: </strong>The proposed model achieves 55.4% to 64.2% satisfaction for doctors and 69-74.5% satisfaction for patients in terms of individual satisfaction. In addition, the proposed BLFCAN can increase overall satisfaction by 26.5-29% in fewer rounds, and it can alter the negotiation strategy in a flexible manner for various negotiation scenarios.</p><p><strong>Conclusions: </strong>BLFCAN reduces communication time and cost, helps avoid potential conflicts, and reduces the impact of emotions and biases on decision-making. In addition, the BLFCAN model improves the agreement satisfaction of both parties and the total social welfare.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"67"},"PeriodicalIF":3.3,"publicationDate":"2025-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11812200/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143390156","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":"Exploration of the optimal deep learning model for english-Japanese machine translation of medical device adverse event terminology.","authors":"Ayako Yagahara, Masahito Uesugi, Hideto Yokoi","doi":"10.1186/s12911-025-02912-0","DOIUrl":"10.1186/s12911-025-02912-0","url":null,"abstract":"<p><strong>Background: </strong>In Japan, reporting of medical device malfunctions and related health problems is mandatory, and efforts are being made to standardize terminology through the Adverse Event Terminology Collection of the Japan Federation of Medical Device Associations (JFMDA). Internationally, the Adverse Event Terminology of the International Medical Device Regulators Forum (IMDRF-AET) provides a standardized terminology collection in English. Mapping between the JFMDA terminology collection and the IMDRF-AET is critical to international harmonization. However, the process of translating the terminology collections from English to Japanese and reconciling them is done manually, resulting in high human workloads and potential inaccuracies.</p><p><strong>Objective: </strong>The purpose of this study is to investigate the optimal machine translation model for the IMDRF-AET into Japanese for the part of a function for the automatic terminology mapping system.</p><p><strong>Methods: </strong>English-Japanese parallel data for IMDRF-AET published by the Ministry of Health, Labor and Welfare in Japan was obtained from 50 sentences randomly extracted from the terms and their definitions. These English sentences were fed into the following machine translation models to produce Japanese translations: mBART50, m2m-100, Google Translation, Multilingual T5, GPT-3, ChatGPT, and GPT-4. The evaluations included the quantitative metrics of BiLingual Evaluation Understudy (BLEU), Character Error Rate (CER), Word Error Rate (WER), Metric for Evaluation of Translation with Explicit ORdering (METEOR), and Bidirectional Encoder Representations from Transformers (BERT) score, as well as qualitative evaluations by four experts.</p><p><strong>Results: </strong>GPT-4 outperformed other models in both the quantitative and qualitative evaluations, with ChatGPT showing the same capability, but with lower quantitative scores, in the qualitative evaluation. Scores of other models, including mBART50 and m2m-100, lagged behind, particularly in the CER and BERT scores.</p><p><strong>Conclusion: </strong>GPT-4's superior performance in translating medical terminology, indicates its potential utility in improving the efficiency of the terminology mapping system.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"66"},"PeriodicalIF":3.3,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11807297/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143373811","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}