JMIR Medical Informatics最新文献

{"title":"An Artificial Intelligence-Based Framework for Predicting Emergency Department Overcrowding: Development and Evaluation Study.","authors":"Orhun Vural, Bunyamin Ozaydin, Khalid Y Aram, James Booth, Brittany Freeman Lindsey, Abdulaziz Ahmed","doi":"10.2196/73960","DOIUrl":"10.2196/73960","url":null,"abstract":"<p><strong>Background: </strong>Emergency department (ED) overcrowding remains a critical challenge, leading to delays in patient care and increased operational strain. Current hospital management strategies often rely on reactive decision-making, addressing congestion only after it occurs. However, effective patient flow management requires early identification of overcrowding risks to allow timely interventions. Machine learning (ML)-based predictive modeling offers a solution by forecasting key patient flow measures, such as waiting count, enabling proactive resource allocation and improved hospital efficiency.</p><p><strong>Objective: </strong>The aim of this study is to develop ML models that predict ED waiting room occupancy (waiting count) at 2 temporal resolutions. The first approach is the hourly prediction model, which estimates the waiting count exactly 6 hours ahead at each prediction time (eg, a 1 PM prediction forecasts 7 PM). The second approach is the daily prediction model, which forecasts the average waiting count for the next 24-hour period (eg, a 5 PM prediction estimates the following day's average). These predictive tools support resource allocation and help mitigate overcrowding by enabling proactive interventions before congestion occurs.</p><p><strong>Methods: </strong>Data from a partner hospital's ED in the southeastern United States were used, integrating internal and external sources. Eleven different ML algorithms, ranging from traditional approaches to deep learning architectures, were systematically trained and evaluated on both hourly and daily predictions to determine the models that achieved the lowest prediction error. Experiments optimized feature combinations, and the best models were tested under high patient volume and across different hours to assess temporal accuracy.</p><p><strong>Results: </strong>The best hourly prediction performance was achieved by time series vision transformer plus (TSiTPlus) with a mean absolute error (MAE) of 4.19 and a mean squared error (MSE) of 29.36. The overall hourly waiting count had a mean of 18.11 and a SD (σ) of 9.77. Prediction accuracy varied by time of day, with the lowest MAE at 11 PM (2.45) and the highest at 8 PM (5.45). Extreme case analysis at (mean + 1σ), (mean + 2σ), and (mean + 3σ) resulted in MAEs of 6.16, 10.16, and 15.59, respectively. For daily predictions, an explainable convolutional neural network plus (XCMPlus) achieved the best results with an MAE of 2.00 and a MSE of 6.64. The daily waiting count had a mean of 18.11 and a SD of 4.51. Both models outperformed traditional forecasting approaches across multiple evaluation metrics.</p><p><strong>Conclusions: </strong>The proposed prediction models effectively forecast ED waiting count at both hourly and daily intervals. The results demonstrate the value of integrating diverse data sources and applying advanced modeling techniques to support proactive resource allocation decisions. The implementation of ","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e73960"},"PeriodicalIF":3.8,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12489414/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145082104","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":"Holistic Influence of Multimodal Medical Crowdfunding Affordances on Charitable Crowdfunding Outcome: Systematic Multimodel Analysis Study.","authors":"Yuxuan Du, Yujin Yang, Zihe Li, Jiaolong Xue","doi":"10.2196/75563","DOIUrl":"10.2196/75563","url":null,"abstract":"<p><strong>Background: </strong>Medical crowdfunding has emerged as a critical tool to alleviate the financial burden of health care costs, particularly in regions where economic disparities limit access to medical treatment. Despite its potential, the success rates of medical crowdfunding projects remain low, with only 9% achieving their fundraising goals in China. Previous research has examined isolated factors influencing success, but a holistic understanding of how multimodal affordances-narrativity, visibility, and progress-collectively impact donor behavior and project outcomes is lacking.</p><p><strong>Objective: </strong>This study aims to investigate how medical crowdfunding affordances, as an integrated system, influence the success of charitable crowdfunding projects. Specifically, it explores the roles of narrativity (textual elements), visibility (visual elements), and progress (dynamic updates) affordances, and how these interact with patient demographics to shape donor engagement and fundraising outcomes.</p><p><strong>Methods: </strong>A multimodal analysis was conducted using 1261 medical crowdfunding projects from the Shuidichou platform in China. Machine learning techniques (eg, sentiment analysis via SnowNLP) and regression models were used to examine textual content, visual elements, and progress updates. Control variables included patient age, gender, and beneficiary type. Hypotheses were tested using both continuous (success ratio) and binary (success indicator) measures of project success. In total, 6 models were constructed to examine the influences of affordances.</p><p><strong>Results: </strong>The study found that narrativity affordances-longer titles (model 1a: P=.04; model 3a: P=.03) and detailed surplus fund descriptions (P=.03)-boosted success, while overly lengthy surplus fund explanations had diminishing returns (P=.005). Disease mentions in titles increased donations (model 1a: P=.01; model 3a: P=.003). A neutral tone in the project plan also improved success (P<.001). For visibility affordances, a moderate number of progress photos maximized project success, while excessive visuals reduced impact (P<.001). Progress affordances followed a similar pattern, with a moderate number of updates enhancing success (P<.001). Critically, when all affordances were considered, only progress update frequency retained a strong inverted U-shaped effect on success (P<.001). Demographics, particularly age, also influenced donations: patients at both ends of the age spectrum received greater support , while middle-aged individuals received less (model 1b: P=.02; model 2b: P=.005; model 3b: P=.02).</p><p><strong>Conclusions: </strong>This study advances medical crowdfunding affordance theory by demonstrating the interconnected effects of narrativity, visibility, and progress affordances on project success. Practically, results highlight the importance of strategically crafted titles, targeted demographic disclosures, and balanced ","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e75563"},"PeriodicalIF":3.8,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12494109/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076939","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":"Interpretable Machine Learning for Predicting Adverse Pregnancy Outcomes in Gestational Diabetes: Retrospective Cohort Study.","authors":"Jiaxi Li, Xiali Liu, Shenyang He, Yan Ren","doi":"10.2196/71539","DOIUrl":"10.2196/71539","url":null,"abstract":"<p><strong>Background: </strong>Gestational diabetes mellitus (GDM) affects over 5% of pregnancies worldwide, elevating risks of type 2 diabetes post partum and complications such as fetal death, miscarriage, and congenital abnormalities. Effective GDM management is essential to balance glycemic control and pregnancy outcomes.</p><p><strong>Objective: </strong>We aim to develop interpretable machine learning models using GDM datasets for predicting adverse pregnancy outcomes and identifying key factors through the Shapley additive explanations (SHAP) algorithm, thus supporting improved maternal and infant health.</p><p><strong>Methods: </strong>Data preprocessing and feature selection were performed, with adaptive synthetic sampling used to address class imbalance. Classification models, including logistic regression, random forest, support vector machine, and extreme gradient boosting, were built and enhanced through the stacking method. Model interpretability was assessed with SHAP to quantify feature contributions.</p><p><strong>Results: </strong>Among 1670 patients, 200 experienced adverse outcomes. The stacking model outperformed individual models, achieving an accuracy of 85.6%, a sensitivity of 57.8%, a specificity of 95.9%, and an area under the receiver operating characteristic curve of 0.82 on the test set. External validation on 159 patients showed a decline in performance (accuracy 83.6%, area under the receiver operating characteristic curve 0.67). SHAP analysis identified gestational age, glucose control, and diagnosis time among the most influential predictors, providing clinically meaningful insights into risk factors. Additionally, detailed SHAP-based visualization revealed the distribution of different feature values and their nonlinear impact on outcomes, as well as interaction effects between features. These interpretable analyses enabled a deeper understanding of individual and combined feature contributions, thereby enhancing clinical assessment capabilities.</p><p><strong>Conclusions: </strong>This study underscores the potential of machine learning in predicting adverse outcomes in GDM, with interpretable features offering valuable clinical insights to enhance pregnancy management and maternal-infant health.</p>","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e71539"},"PeriodicalIF":3.8,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12441465/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076971","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":"Real-Time Estimation of Arterial Partial Pressure of Carbon Dioxide in Patients Undergoing General Anesthesia: Predictive Modeling Study.","authors":"Ah Ra Lee, Jun Ho Lee, Sooyoung Yoo, Ho-Young Lee, Hyun Ho Kim","doi":"10.2196/64855","DOIUrl":"10.2196/64855","url":null,"abstract":"<p><strong>Background: </strong>Adequate ventilation in mechanically ventilated patients is contingent upon the monitoring of the arterial partial pressure of carbon dioxide (PaCO2) during general anesthesia. Despite its significance, continuous monitoring remains challenging due to the imprecision of noninvasive estimations and the invasive nature of traditional methods such as arterial blood gas analysis.</p><p><strong>Objective: </strong>This study aimed to develop a machine learning model to continuously estimate PaCO2 in mechanically ventilated patients, with the goal of potentially improving intraoperative monitoring accuracy under general anesthesia.</p><p><strong>Methods: </strong>This retrospective study used the VitalDB dataset from Seoul National University Hospital, comprising records of 6388 noncardiac surgery patients between August 2016 and June 2017. After applying inclusion and exclusion criteria, data from 2304 surgical cases (4651 PaCO2 measurement event points) were analyzed. The CatBoost regressor model was trained to predict PaCO2 using noninvasive physiological parameters and clinical information. The model's performance was evaluated using nested cross-validation across hypocapnic (<35 mm Hg), normocapnic (35-45 mm Hg), and hypercapnic (>45 mm Hg) subgroups and compared to conventional estimation methods based on end-tidal carbon dioxide (ETCO2).</p><p><strong>Results: </strong>The developed model demonstrated superior overall performance compared to traditional estimations. It achieved a mean absolute error of 2.38 mm Hg and an average intraclass correlation coefficient of 0.87. Furthermore, 90.02% of the model's estimations fell within the clinically highly acceptable range (error<±5 mm Hg) while only 1.20% of errors exceeded ±10 mm Hg. Performance improvements were observed across all PaCO2 subgroups.</p><p><strong>Conclusions: </strong>The developed model provides more accurate and reliable estimates of PaCO2 than traditional ETCO2-based methods. This approach shows potential for facilitating real-time monitoring and timely clinical interventions. This study demonstrated the potential of artificial intelligence to enhance continuous monitoring of PaCO2; however, further validation, including prospective studies assessing clinical impact, is necessary.</p>","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e64855"},"PeriodicalIF":3.8,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12439857/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076948","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":"Using Wearable Device and Machine Learning to Predict Mood Symptoms in Bipolar Disorder: Development and Usability Study.","authors":"Chia-Tung Wu, Ming H Hsieh, I-Ming Chen, Lian-Yin Jhao, Ding-Shan Liu, Ssu-Ming Wang, Chia-Ting Wu, Yi-Ling Chien","doi":"10.2196/66277","DOIUrl":"10.2196/66277","url":null,"abstract":"<p><strong>Background: </strong>Bipolar disorder (BD) is a highly recurrent disorder. Early detection, early intervention, and prevention of recurrent bipolar mood symptoms are key to a better prognosis.</p><p><strong>Objective: </strong>This study aims to build prediction models for BD with machine learning algorithms.</p><p><strong>Methods: </strong>This study recruited 24 participants with BD. The Beck Depression Inventory and Young Mania Rating Scale were used to evaluate depressive and manic episodes, respectively. Using digital biomarkers collected from wearable devices as input, 6 machine learning algorithms (logistic regression, decision tree, k-nearest neighbors, random forest, adaptive boosting, and Extreme Gradient Boosting) were used to build predictive models.</p><p><strong>Results: </strong>The prediction model for depressive symptoms achieved 83% accuracy, an area under the receiver operating characteristic curve (AUROC) of 0.89, and an F1-score of 0.65 on testing data. The prediction model for manic symptoms achieved 91% accuracy, an AUROC of 0.88, and an F1-score of 0.25 on testing data. With the interpretable model Shapley Additive Explanations, we found that relatively high resting heart rate, low activity, and lack of sleep may predict depressive symptoms.</p><p><strong>Conclusions: </strong>This study demonstrated that digital biomarkers could be used to predict depressive and manic symptoms. This prediction model may be beneficial for the early detection of mood symptoms, facilitating timely treatment and helping to prevent BD recurrence.</p>","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e66277"},"PeriodicalIF":3.8,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12440259/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076946","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":"Evidence for the Use of Patient-Reported Outcome Measures in the Treatment of Patients With Noncommunicable Diseases: Systematic Review.","authors":"Marie Villumsen, Benedikte Irene von Osmanski, Kirsten Elisabeth Lomborg, Kirstine Skov Benthien","doi":"10.2196/66160","DOIUrl":"10.2196/66160","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;The use of patient-reported outcome measures (PROMs) as a clinical tool for screening and decision-making has gained widespread interest, with numerous implementation activities across specialties, even though the evidence has not been clear until now.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;The aim of this study was to assess the evidence for using PROMs in clinical practice for patients with diabetes, chronic obstructive pulmonary disease (COPD), heart disease, rheumatoid arthritis (RA), and inflammatory bowel disease (IBD). Additionally, we sought to determine the characteristics of the most effective PROM interventions.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We conducted a systematic review of published randomized controlled trials (RCTs) on the use of PROMs for clinical purposes, such as systematic PROM assessment alone or with a predefined PROM-based decision-making method. Eligible studies included adult patients (&gt;18 years) with diabetes, COPD, heart disease, RA, or IBD. We excluded studies using PROMs as an outcome measure or otherwise not meeting the inclusion criteria. We searched the PubMed/MEDLINE, CINAHL, EMBASE, and Web of Science databases until February 2023. Two investigators independently screened titles, abstracts, and relevant full texts. Three investigators completed data extraction and risk-of-bias assessment using version 2 of the Cochrane risk-of-bias tool for randomized trials (RoB 2). The data were presented in a narrative synthesis and in summarized form.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;The search yielded 21,203 papers, 686 (3.2%) full-text papers were screened, and 56 (8.2%) original studies were included in the review. The studies included patients with heart disease (n=17, 30.4%), COPD (n=13, 23.2%), diabetes (n=10, 17.9%), IBD (n=9, 16.1%), and RA (n=6, 10.7%), as well as patients with mixed diagnoses (n=1, 1.8%). All interventions incorporated systematic PROM assessments. Some interventions additionally used a predefined method for PROM-based decision-making (n=19, 33.9%) or PROM-based dialogue (n=9, 16.1%), while 5 (8.9%) interventions aimed to substitute face-to-face consultations. The predominant mode of PROM administration was over the phone, followed by electronic devices and apps. Endpoints included disease activity, health care use, mortality, mental well-being, quality of life, self-efficacy, self-care, daily functioning, and other outcomes. Six studies with a low risk of bias demonstrated a positive effect or noninferiority of the PROM intervention.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;The evidence base for clinical use of PROMs is sparse, with few studies evaluated to have a low or a medium risk of bias. The clinical use of PROMs does not appear superior to usual care in the five included chronic diseases on any endpoint. To guide further research, we highlighted 6 (10.7%) studies with a low risk of bias and PROM interventions with a positive effect. These were characterized by symptom ","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e66160"},"PeriodicalIF":3.8,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12485264/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076959","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":"Clinical Trial Schedule of Activities Specification using FHIR Definitional Resources.","authors":"Andrew Richardson, Patrick Genyn","doi":"10.2196/71430","DOIUrl":"https://doi.org/10.2196/71430","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Clinical research studies rely on schedules of activities (SoAs) to define what data must be collected and when. Traditionally presented in tabular form within study protocols, SoAs are critical for ensuring data quality, regulatory compliance, and correct study execution. Recent efforts, such as the HL7 Vulcan Schedule of Activities Implementation Guide (SoA IG), have introduced Fast Healthcare Interoperability Resources (FHIR) as a standard for representing SoAs digitally. However, current approaches primarily handle simple schedules and do not adequately capture complex requirements such as conditional branching, repeat cycles, or unscheduled events-features essential for many study designs, particularly in oncology.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;This work aimed to extend SoA representation methods to address these limitations. Specific objectives were: (1) to develop methods for defining multiple SoA paths within a single model; (2) to specify conditional scheduling requirements; (3) to design a human-readable syntax for study specifications; (4) to reflect these requirements as FHIR definitional resources; and (5) to test bidirectional conversion between graph-based SoA models and FHIR representations.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;Building on prior work, SoAs were modeled using directed graphs in which nodes represented interactions (e.g., visits) or activities, and edges defined transitions. Attributes were added to capture timing, conditional rules, and repeatability. Graph-based models were translated into FHIR PlanDefinitions and related resources (ActivityDefinition, ResearchStudy, ResearchSubject). Extensions to PlanDefinition were developed (soaTimePoint and soaTransition) to store graph-specific attributes. Proof-of-concept models were implemented and tested using Python, NetworkX, pandas, and FHIR Shorthand, with validation conducted through FHIR servers to ensure structural equivalence and information retention.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;The graph-based approach successfully modeled multiple paths, unscheduled events, and conditional rules within a single SoA. Edge attributes such as transitionDelay and transitionRule enabled accurate timing calculations and runtime evaluation of permitted paths. Conditional scheduling was expressed using a parameterized syntax interpretable by logic engines. More than 25 study protocols of varying complexity were tested; all could be represented without information loss. The proposed FHIR extensions allowed PlanDefinition resources to fully capture SoA graphs rather than limited tabular forms. Round-trip testing confirmed that graph models and FHIR resources could be converted without loss of fidelity. The approach also highlighted inconsistencies in some protocol specifications, suggesting its utility for protocol quality assurance.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;This work demonstrates that graph-based modeling, combined with targeted FHIR PlanDef","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":" ","pages":""},"PeriodicalIF":3.8,"publicationDate":"2025-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076885","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":"Assessing Internet Quality Across Public Health Centers in Indonesia: Cross-Sectional Evaluation Study.","authors":"Dewi Nur Aisyah, Agus Heri Setiawan, Alfiano Fawwaz Lokopessy, Chyntia Aryanti Mayadewi, M Thoriqul Aziz Endryantoro, Viktor Wibowo, Sarah Disviana, Indra Laksana, Mohammad Aviandito, Zisis Kozlakidis, Logan Manikam","doi":"10.2196/65940","DOIUrl":"10.2196/65940","url":null,"abstract":"<p><strong>Background: </strong>Primary health care centers (Puskesmas) serve as the cornerstone of Indonesia's health care system, providing integrated services aimed at improving individual health through prevention, treatment, and health promotion. To fulfill these roles effectively, robust technological infrastructure, particularly reliable internet connectivity, is increasingly essential. Assessing the availability and quality of internet access in Puskesmas is therefore a critical step in understanding their readiness to implement digital health initiatives and fulfill their responsibilities in delivering accessible and effective healthcare services.</p><p><strong>Objective: </strong>This study provides a national baseline assessment of internet quality and its relevant IT infrastructure in more than 10,000 Puskesmas across Indonesia.</p><p><strong>Methods: </strong>A cross-sectional survey was conducted throughout all Puskesmas (10,382) in 34 provinces in Indonesia, using an online questionnaire. Categorization was done to analyze internet quality level results.</p><p><strong>Results: </strong>A total of 10,378 out of 10,382 public health centers (99.96%) participated in this study. Overall, 745 of 10,382 (7.18%) did not have internet access, 1487 (14.33%) had limited internet access, 5567 (53.64%) had sufficient internet access, and 2579 (24.85%) had sufficient and fast internet access. Moreover, 832 of 10,382 Puskesmas (8.02%) did not have 24-hour electricity, 44,196 (43.7%) had a central processing unit (CPU) with i3 specifications, 43,044 (42.56%) had 512 GB hard disk capacity, and 67,272 (66.5%) used antivirus.</p><p><strong>Conclusions: </strong>Although 79% (8201/10,382) of Puskesmas in Indonesia already had sufficient internet access, 21% (2180/10,382) still have limited and insufficient access. To ensure universal internet availability, it is essential to build collaborative support among internet providers and government to foster the availability and use of internet satellites, high-quality computers, and electrical power to support internet connectivity.</p>","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e65940"},"PeriodicalIF":3.8,"publicationDate":"2025-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12435787/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145071068","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":"Performance of Natural Language Processing for Information Extraction From Electronic Health Records Within Cancer: Systematic Review.","authors":"Simon Dahl, Martin Bøgsted, Tomer Sagi, Charles Vesteghem","doi":"10.2196/68707","DOIUrl":"10.2196/68707","url":null,"abstract":"<p><strong>Background: </strong>Over the last decade, natural language processing (NLP) has provided various solutions for information extraction (IE) from textual clinical data. In recent years, the use of NLP in cancer research has gained considerable attention, with numerous studies exploring the effectiveness of various NLP techniques for identifying and extracting cancer-related entities from clinical text data.</p><p><strong>Objective: </strong>We aimed to summarize the performance differences between various NLP models for IE within the context of cancer to provide an overview of the relative performance of existing models.</p><p><strong>Methods: </strong>This systematic literature review was conducted using 3 databases (PubMed, Scopus, and Web of Science) to search for articles extracting cancer-related entities from clinical texts. In total, 33 articles were eligible for inclusion. We extracted NLP models and their performance by F1-scores. Each model was categorized into the following categories: rule-based, traditional machine learning, conditional random field-based, neural network, and bidirectional transformer (BT). The average of the performance difference for each combination of categorizations was calculated across all articles.</p><p><strong>Results: </strong>The articles covered various scenarios, with the best performance for each article ranging from 0.355 to 0.985 in F1-score. Examining the overall relative performances, the BT category outperformed every other category (average F1-score between 0.2335 and 0.0439). The percentage of articles on implementing BTs has increased over the years.</p><p><strong>Conclusions: </strong>NLP has demonstrated the ability to identify and extract cancer-related entities from unstructured textual data. Generally, more advanced models outperform less advanced ones. The BT category performed the best.</p>","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e68707"},"PeriodicalIF":3.8,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12431712/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145056539","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":"Lessons Learned From Building a Data Platform for Longitudinal, Analytical Use Cases and Scaling to 77 German Hospitals: Implementation Report.","authors":"Markus Bockhacker, Peter Martens, Clara von Münchow, Sarah Löser, Rosita Günther, Ralf Kuhlen, Olaf Kannt, Sebastian Ortleb","doi":"10.2196/69853","DOIUrl":"10.2196/69853","url":null,"abstract":"<p><strong>Background: </strong>Increasing adoption of electronic health records (EHRs) enables research on real-world data. In Germany, this has been limited to university hospitals, and data from acute care hospitals below the university level are lacking. To address this, we used established design patterns to build a data platform that aggregates and standardizes pseudonymized EHR data with patients' consent.</p><p><strong>Objective: </strong>We report on the design and implementation of the research platform, as well as patient participation and lessons learned during the scaling of the platform, to incorporate real-world data (with participant consent) from 77 hospitals into a unified data lake.</p><p><strong>Methods: </strong>Due to variations in EHR adoption, IT infrastructure, software vendors, interface availability, and regulatory requirements, we used an agile development cycle that involves constant, incremental standardization of data. We implemented a layered lambda infrastructure built on Apache Hadoop. Decentralized connectors ensured data minimization and pseudonymization.</p><p><strong>Unlabelled: </strong>We successfully scaled our data model both vertically and horizontally in 77 hospitals. However, we encountered issues during the scaling of real-time data pipelines and IHE (Integrating the Healthcare Enterprise) interfaces. During the first 2 years, patients were asked to consent to secondary data use 1,475,244 times during inpatient admission. We registered 1,023,633 broad instances of consent (consent rate 70.2%).</p><p><strong>Conclusions: </strong>Patients are generally willing to provide consent for secondary use of their data, but obtaining consent requires considerable effort. Building a research data platform is not an end goal, but rather a necessary step in collecting and standardizing longitudinal data to enable research on real-world data. Through the combination of agile development, phased rollouts, and very high levels of automation, we have been able to achieve fast turnaround times for incorporating user feedback and are constantly improving data quality and standardization.</p>","PeriodicalId":56334,"journal":{"name":"JMIR Medical Informatics","volume":"13 ","pages":"e69853"},"PeriodicalIF":3.8,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12431789/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145056597","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}