Health Informatics Journal最新文献

Objective: This study presents a structured evaluation of large language models (LLMs) in predicting suicide methods based exclusively on indirect forensic psychiatric indicators. Methods: Ninety-two forensic psychiatric cases (2019-2024), involving survivors of suicide attempts formally examined in medico-legal contexts, were retrospectively analyzed. Variables included age, sex, psychiatric diagnosis, previous suicide attempts, psychiatric medication use, impulsivity, and consciousness at emergency admission. Six LLMs were tested: ChatGPT-4o, ChatGPT-4o Mini, ChatGPT-O3 (OpenAI), Gemini 2.0 Flash, Gemini 2.5 Pro, and Gemini 2.5 Flash (Google DeepMind). Each case was converted into a standardized anonymized prompt. Model predictions were categorized by blinded forensic physicians and evaluated using accuracy, precision, recall, F1-score, and Cohen's Kappa for 1-month reproducibility. Results: Gemini 2.5 Flash achieved the highest performance with 76.09% accuracy, 46.9% F1-score, and 45.2% recall. It accurately predicted the dominant method, medication overdose, but underperformed for rare categories. Temporal reproducibility was moderate (κ = 0.582), while other models exhibited lower and less stable performance. Conclusion: LLMs can infer suicide methods from indirect psychiatric data with encouraging accuracy. However, limitations in detecting rare methods and maintaining temporal consistency suggest the need for further methodological refinement and external validation prior to forensic application.

Objectives: Addressing data duplication is one of the most important issues in electronic health record (EHR) processing since the nature of data collection in the field. It does not only affect the data quality in healthcare management, but also the reliability in the downstream analyses. In this paper, we propose a comprehensive data de-duplication framework tailored for medical databases to tackle data duplication for a kidney disease identification, Acute Kidney Failure (AKF). Methods: The proposed work begins with the data joining from various sources, basic data de-duplication which automatically removes the dirty texts, medical note-event extraction since the data could be sources for further de-duplication, NLP data de-duplication based on a pre-trained model, data mapping for integration, unrelated data and outlier elimination, and eventually data imputation by a clustered based imputer. Results: We illustrated our de-duplication framework on MIMIC-III database both on the de-duplication task and the classification task based on AKF. The experiments demonstrated that the proposed work could achieve up to 99.59% accuracy or 23% higher than the traditional method and could achieve a high classification accuracy at 86 % and the F1-score at 0.87, which outperformed the traditional method, and the original dataset without any modification. Conclusion: These results demonstrated that the framework can potentially address the data duplication issue in healthcare effectively.

Background: The rapid expansion of digital health services (DHS) highlights the need to assess their accessibility and effectiveness, particularly among older adults. Despite increasing digitalization, many older individuals still face barriers, including limitations in digital competence and access. Objective: This study examines the use, barriers, and perceived benefits of DHS among individuals aged 75 and older in Finland. Methods: A nationwide survey was conducted in March 2023 using both electronic and paper questionnaires. In addition to descriptive analysis, regression analysis was performed to identify variables associated with perceived benefits of digital health services. Results: Of the 1124 responses (1011 electronic, 113 paper), 1100 were fully completed. Overall, 84% of respondents had used DHS, with usage being higher among those under 85 years (87%) than those over 85 (57%). The majority of respondents (82%) reported using the national Omakanta service, which grants access to personal health information. Digital competence and the number of services used were the strongest predictors of perceived benefits, alongside higher satisfaction, service frequency, and female gender. Conclusions: DHS adoption among older adults, especially in Finland, may be higher than previously reported. However, digital social services remain underdeveloped. Addressing the digital divide is essential to ensuring equitable access.

Objective: Women with prior hypertensive disorders of pregnancy (HDP) have increased risk of developing future cardiovascular disease. The objective of this scoping review was to map the literature regarding the use of eHealth measures in cardiovascular follow-up after HDP and identify research gaps. Methods: A systematic search was conducted in four databases. Primary research articles and guidelines were included. Abstract screening, full-text assessment and data extraction was performed to summarize the findings. Results: The search identified 4830 articles and 12 guidelines. Eleven publications and one guideline were included in the analyses. Various eHealth interventions were assessed, such as remote blood pressure monitoring, physical activity and weight management, with follow-up time from 6 weeks to 4 years. eHealth interventions targeting blood pressure and physical activity showed statistically significant positive effects. Conclusion: The scoping review identified eHealth interventions for cardiovascular follow-up after HDP that may empower women to optimize their cardiovascular health.

Suicide is a critical public health issue worldwide, influenced by environmental factors such as economic stress and limited social support, as well as individual risk factors. Patients with chronic health conditions may face heightened vulnerability due to overlapping psychological and medical challenges. This research explores the application of Machine Learning (ML) techniques to identify suicide risk among such patients, utilizing data from the National Health and Nutrition Examination Survey (NHANES). The study incorporated demographic, clinical, and psycho-social variables, including depression, substance use, hypertension, and diabetes, to develop predictive models. Several ML algorithms were trained and evaluated using standard performance metrics to assess predictive accuracy. Among the models, Gradient Boosting Machine (GBM) achieved the strongest performance, with a receiver operating characteristic area under the curve (ROC-AUC) of 0.9479. Random Forest also performed exceptionally, with a ROC-AUC of 0.9301, while four additional models showed competitive results. These algorithms effectively captured complex nonlinear relationships and interactions between multiple risk factors, demonstrating their suitability for multivariable health data. The findings underscore the potential of integrating ML into Electronic Medical Records (EMRs) as decision-support tools to identify high-risk patients. Early detection enables timely interventions, which may significantly improve mental health outcomes and reduce suicide risk.

Objective: This study aimed to investigate how nurses' backgrounds, documentation skills, and experiences with documentation practices and information systems usage influence documentation hazards and to determine whether these hazards are linked to technology-induced errors (TIEs). Methods: An online survey was conducted to collect data from 3065 registered nurses working in Finnish hospitals and in acute, primary, and home care services regarding their experiences with electronic health records (EHRs) or client information systems (CIS). The data were analysed using linear and logistic multilevel models to identify patterns and correlations. Results: User interaction with EHR/CIS systems significantly influenced documentation hazards across different work environments. Perceived system-provided documentation support and documentation hazards were identified as contributors to TIEs. Conclusions: Improving system design and documentation support is a desirable goal, but it is not sufficient to mitigate documentation hazards and promote efficient practices. To achieve the best possible results, skilled users are needed to operate these systems.

Background: The introduction of the RTS, S (Mosquirix®) malaria vaccine in Cameroon represents a significant public health milestone. This study analyzed social media sentiment surrounding the vaccine rollout using natural language processing and machine learning. Methods: Data from Twitter (X) and Facebook (Meta) regarding the RTS, S vaccine in Cameroon was analyzed using the Hugging Face Transformer library for sentiment evaluation. The data was pre-processed, cleaned, and visualized with Matplotlib. Results: The sentiment analysis revealed that 42.0% of reactions were negative, 40.0% were positive, and 18.0% were neutral, indicating a nearly even split between skeptical and supportive viewpoints among Cameroonian users regarding the vaccine rollout. Conclusion: The research highlights the necessity for targeted communication strategies to address public concerns and foster vaccine confidence. Sentiment analysis can act as a real-time tool, offering policymakers valuable insights into public reactions and attitudes toward immunization and other health initiatives. These findings reveal significant public skepticism that must be addressed through evidence-based communication strategies focused on vaccine safety, efficacy data from pilot programs, and engagement with community leaders to counter misinformation.

According to the Immunization Status Survey conducted by the Tennessee Department of Health in 2023, Tennessee ranks in the bottom 25th percentile among states for vaccination rates by the age of 24 months, based on the full series of recommended vaccines. To tackle this issue, the SmartSHOTS mobile application (mobile app) was developed to reduce vaccination barriers for children aged 0-24 months. The mobile app includes vaccine information, the ability to add and calculate vaccine due dates, and locating health departments and transportation services based on zip codes. The mobile app was developed and usability tested using an iterative design process, based on a needs assessment conducted across regions of Tennessee with community members who served on the state's county health council. These community members also reviewed the mobile app wireframes. Parents or guardians of children aged 0-24 months living in Tennessee evaluated the usability of the mobile app.

Clinical Nursing Information Systems (CNISs) and Standardized Nursing Terminologies (SNTs) significantly enhance the quality of care, promote interoperability, and enable measurable nursing outcomes. However, their adoption remains inconsistent, particularly in low- and middle-income countries (LMICs). This commentary reframes the existing gap as an issue of equity and systems design while providing a feasibility-prioritized roadmap tailored for LMICs. The article supports a sequenced approach that distinguishes between short-term actions and longer-term initiatives. Short-term actions include stabilizing infrastructure, developing open-source CNIS models, initiating terminology localization pilots, implementing essential data privacy safeguards, and providing targeted in-service training. In contrast, longer-term initiatives involve establishing national standards and exchanges, securing sustainable financing, cultivating leadership pipelines and curricula, and promoting cross-border interoperability and evaluation. Furthermore, it delineates various financing mechanisms-including concessional loans, performance-based grants, and collective procurement-while also addressing strategic considerations related to policy and governance frameworks. The commentary concludes with an explicit call to action: policymakers, donors, nursing leaders, educators, and vendors must collaborate to integrate structured nursing data into routine care and national platforms. Bridging this gap will render nursing work more visible, enhance decision support, and foster learning health systems within hospitals and communities worldwide.

Big Data in Internet of Healthcare Things (IoHT) environments includes large volumes of structured and unstructured clinical information. The Hadoop Distributed File System (HDFS) is widely used for its scalability and ability to run on commodity hardware. However, it offers limited native encryption, leaving data vulnerable to security risks. Although several encryption techniques exist, traditional algorithms still face performance and security limitations with large-scale medical datasets. Therefore, this study introduces a hybrid encryption framework designed to enhance security in IoHT environments that process large-scale medical Big Data. The framework combines Attribute-Based Encryption (ABE) with the Blowfish cipher to secure data generated by heterogeneous medical devices across the IoHT infrastructure. The proposed approach is benchmarked against established hybrid schemes-CP-ABE + HE, HE + BF, and CP-ABE + AES-to provide a comparative assessment of its security strength and computational performance. The performance assessment employed key computational metrics, including system efficiency, encryption latency, and decryption latency. Experimental results demonstrate that the proposed hybrid scheme delivers superior performance compared to existing approaches, attaining a peak efficiency of 98.5%. The method further achieved encryption and decryption times of 6.8 min and 5.7 min, respectively, indicating improved computational handling of large-scale IoHT data.