DIGITAL HEALTH最新文献

{"title":"How do perceived AI algorithmic recommendation characteristics influence health information adoption? A PLS-SEM and fsQCA analysis.","authors":"Wei Zhu, Ping Ouyang, Shuqin Li, Jieyuan Lan, Xue Ke","doi":"10.1177/20552076261484764","DOIUrl":"10.1177/20552076261484764","url":null,"abstract":"<p><strong>Objective: </strong>This study examines how users' perceptions of AI algorithmic recommendation characteristics are associated with health information adoption intention on online healthcare platforms. It further investigates the mediating roles of trust and perceived usefulness in this process.</p><p><strong>Methods: </strong>A cross-sectional online survey was conducted among users of Haodf.com, a major online healthcare platform in China. After a pilot test, 310 valid responses were collected. Partial least squares structural equation modeling (PLS-SEM) was used to examine the hypothesized relationships, while fuzzy-set qualitative comparative analysis (fsQCA) was employed to identify multiple configurations associated with high health information adoption intention.</p><p><strong>Results: </strong>Perceived AI algorithmic recommendation characteristics were positively associated with both trust and perceived usefulness. Trust was positively associated with perceived usefulness and health information adoption intention, while perceived usefulness showed a stronger direct association with adoption intention (<i>β</i> = 0.52, <i>p</i> < 0.001). Mediation analysis indicated that these recommendation characteristics were indirectly associated with adoption intention through trust, perceived usefulness, and the theoretically specified pathway linking trust and perceived usefulness. fsQCA further identified multiple configurations associated with high health information adoption intention.</p><p><strong>Conclusion: </strong>Users' adoption intention toward AI-recommended health information is jointly associated with perceived recommendation characteristics and cognitive evaluations. Trust represents a credibility-based evaluation, whereas perceived usefulness appears to be more proximally related to adoption intention. These findings provide insights for optimizing algorithmic recommendation strategies in digital health platforms.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261484764"},"PeriodicalIF":3.9,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13527329/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148868012","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":"Design and multidimensional expert evaluation of a national health information portal-integrated generative AI chatbot for health literacy and awareness.","authors":"Hlib Aleksandrenko, Martin Komenda","doi":"10.1177/20552076261481326","DOIUrl":"10.1177/20552076261481326","url":null,"abstract":"<p><strong>Background: </strong>Generative AI (GenAI) shows promise for health literacy and awareness, with large language models (LLMs) increasingly used for health information-seeking despite concerns about misinformation. National health information portals provide authoritative, evidence-based content but lack conversational interfaces. Retrieval-augmented generation (RAG) offers a potential solution by grounding responses in verified health information portal content, yet evidence on such integration remains limited.</p><p><strong>Objective: </strong>This study aimed to design a national health information portal-integrated generative AI chatbot to support access to trustworthy health information and to evaluate its response characteristics relative to general-purpose LLMs through a multidimensional expert assessment.</p><p><strong>Methods: </strong>A RAG-enhanced chatbot (NZIP AI assistant) was designed for the Czech Republic's National Health Information Portal and evaluated by eleven experts who compared it with ChatGPT, Microsoft Copilot, and Google Gemini across 67 health-related questions. Responses were rated on trustworthiness, clarity, user-friendliness, and relevance using a 5-point Likert scale (1 = best, 5 = worst). Content analysis examined response characteristics.</p><p><strong>Results: </strong>In total, 267 assessments were collected. NZIP AI assistant achieved the highest trustworthiness score (mean 1.21) but ranked lowest on clarity (1.53), user-friendliness (1.46), and relevance (1.65). General-purpose LLMs demonstrated the inverse pattern. Content analysis revealed that the NZIP AI assistant provided source citations in 100% of responses, compared with 0-3% for general-purpose LLMs; included explicit safety disclaimers in 85% of responses, compared with 5-18%; and declined 10.4% of queries outside its knowledge base.</p><p><strong>Conclusions: </strong>National health information portals can integrate generative AI with RAG architecture to deliver verified, source-grounded health information, though trade-offs in user-oriented dimensions warrant consideration in future implementations. Still, this expert evaluation assessed response characteristics rather than direct effects on health literacy, awareness, or health outcomes.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261481326"},"PeriodicalIF":3.9,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13527369/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148867919","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":"Potential legacy effect of digital therapeutics on mortality in CHD patients treated with PCI: 8-Year result of BAMA trial.","authors":"Lei Dong, Yuxi Li, Guopeng Zhou, Jin Fang, Jie Jiang, Fangfang Fan, Jia Jia, Yan Zhang, Bo Zheng, Yimei Zheng, Long Zhang, Yan Zhang, Jianping Li","doi":"10.1177/20552076261485113","DOIUrl":"10.1177/20552076261485113","url":null,"abstract":"<p><strong>Background: </strong>The original BAMA randomized trial showed that a mobile app-based self-management digital therapeutics (DTx) intervention improved medication adherence and cardiovascular risk-factor control at 12 months in patients with coronary heart disease (CHD). Whether these early benefits are associated with favorable long-term outcomes-a potential legacy effect-remains uncertain.</p><p><strong>Methods: </strong>The BAMA pilot trial was a parallel-group, open-label, single-center randomized controlled trial in which 300 patients with CHD were assigned 1:1 to regular follow-up with or without DTx. The present study was a 12-month landmark post-trial analysis with follow-up through June 2025, corresponding to up to 8 years after randomization. Participants in the landmark cohort were analyzed according to original randomized assignment. Cox models evaluated all-cause and cardiovascular mortality, with Firth penalized Cox regression as a sparse-event sensitivity analysis. Longitudinal lipid trajectories were assessed using linear mixed-effects models based on actual measurement timing.</p><p><strong>Results: </strong>Among 290 participants who completed the 12-month visit, 289 had ascertainable post-landmark follow-up. During a median post-trial follow-up of 7.1 years, 22 participants died, including 13 from cardiovascular causes. After adjustment for baseline age and sex, original DTx assignment was associated with lower all-cause mortality (hazard ratio [HR], 0.38; 95% CI, 0.15-0.98; P=.045) and cardiovascular mortality (HR, 0.19; 95% CI, 0.04-0.87; P=.032). Firth penalized analyses yielded directionally consistent results. No significant group-by-time interactions were observed for total cholesterol, triglycerides, high-density lipoprotein cholesterol, or low-density lipoprotein cholesterol.</p><p><strong>Conclusions: </strong>Original DTx assignment was associated with lower post-landmark all-cause and cardiovascular mortality despite broadly comparable longitudinal lipid trajectories. These findings may be consistent with a potential legacy effect and warrant confirmation in larger prospective studies.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261485113"},"PeriodicalIF":3.9,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13527337/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148868023","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":"How telemonitoring impacts the responsibilities of patients and healthcare professionals in cardiology: A Q-methodology study.","authors":"Eva Meier, Marieke A R Bak, Job van Exel, Michiel M Winter, Mariette van den Hoven, Tessel Rigter, Marlies P Schijven","doi":"10.1177/20552076261484036","DOIUrl":"10.1177/20552076261484036","url":null,"abstract":"<p><strong>Objective: </strong>Telemonitoring is widely used in cardiology but raises questions about the impact on roles and responsibilities for patients and caregivers. This study explores this using a Dutch cardiology telemonitoring program.</p><p><strong>Methods: </strong>Q-methodology was employed, combining quantitative and qualitative data from semi-structured interviews. Participants ranked statements on caregiver and patient roles and responsibilities in telemonitoring and explained their rankings. By-person factor analysis was used to identify clusters in the statement rankings. These factors were interpreted using interview data.</p><p><strong>Results: </strong>Thirty participants revealed four viewpoints on roles and responsibilities in telemonitoring: (1) <i>Balancing Act</i>, (2) <i>From Clinical Observers to Remote Monitors</i>, (3) <i>Isolated Responsibility</i>, and (4) <i>Responsible but not Alone</i>. All viewpoints agree on three aspects: patients are responsible for taking and submitting accurate measurements, that nurses' roles are shifting, and that technology developers are accountable for device reliability while caregivers are responsible for data-usage in decision-making. Disagreements between viewpoints remain on four aspects: the extent of patients' responsibility regarding understanding measurements and alarming caregivers, changes in nurses' decision-making responsibilities, and how caregiving responsibilities are shared.</p><p><strong>Conclusion: </strong>Ambiguity about responsibilities create practical and ethical challenges, including increased workload for nurses, lack of overview among cardiologists, anxiety in patients, and inconsistent care delivery. Unrecognized or unaddressed differences in views about roles and responsibilities using telemonitoring can negatively affect patient care. Appropriate implementation and use of telemonitoring requires sound ethical prologue and dialogue, not only to sensitize stakeholders about differing interpretations of responsibility, but also to ensure a structured approach when implementing telemonitoring to support best care.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261484036"},"PeriodicalIF":3.9,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13527333/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148868033","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 scoping review of how automation, self-tracking dose characteristics, and participant age are operationalized in digital physical activity interventions.","authors":"Jingchuan Wu, Alexa Maiorano, Scherezade K Mama, Jonathan G Hakun, David E Conroy","doi":"10.1177/20552076261475799","DOIUrl":"10.1177/20552076261475799","url":null,"abstract":"<p><strong>Background: </strong>Self-tracking physical activity can involve receiving behavioral feedback from a device, engaging in manual self-monitoring, or a combination of the two.</p><p><strong>Objective: </strong>This scoping review examines the implementation of digital self-tracking interventions aimed at promoting physical activity among adults, emphasizing automation formats, dosage metrics, and age-specific adaptations.</p><p><strong>Methods: </strong>A scoping review was conducted of randomized intervention studies published between 2007 and 2025. Eligible studies examined digital self-tracking interventions targeting physical activity in adult populations. Data were charted to summarize intervention characteristics, including self-tracking format, automation level, intervention duration, tracking frequency, and age-related patterns.</p><p><strong>Results: </strong>A total of 197 studies involving 38,492 participants were included. Semi-automated approaches were used most frequently (52%), followed by automated (31%) and manual (7%) methods, and multiple-method approaches (10%). Wearable devices were the most common intervention format (37%), often combined with other digital applications (33%). Intervention duration averaged 21 weeks, and daily self-tracking was the most frequently reported tracking schedule (60%). Few age-related differences in self-tracking format and frequency were observed; however, interventions targeting younger adults tended to be shorter in duration, whereas those targeting middle-aged and older adults more often involved extended self-tracking periods.</p><p><strong>Conclusions: </strong>Digital self-tracking interventions for physical activity in adults show substantial variability in format and dosage. The findings highlight the need for age-responsive and dosage-specific strategies to improve engagement and support physical activity behavior change.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261475799"},"PeriodicalIF":3.9,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13527340/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148867844","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":"Barriers to artificial intelligence chatbot utilization among patients near the Iran-Afghanistan border: A qualitative interview study.","authors":"Mohsen Khosravi, Fatemezahra Zahrayi, Fatemeh Beiky, Ali Miri, Motahareh Nakhaei, Djavad Ghoddoosi-Nejad, Saeede Saadatjoo","doi":"10.1177/20552076261485226","DOIUrl":"10.1177/20552076261485226","url":null,"abstract":"<p><strong>Objectives: </strong>Artificial Intelligence (AI) presents significant advantages in healthcare. The Iran-Afghanistan border region, marked by low socioeconomic status, experiences poor health outcomes, especially among women and children. The presence of armed conflict in the region has further exacerbated the situation. This study examined barriers to patient adoption of AI chatbots in this region.</p><p><strong>Methods: </strong>This was a qualitative exploratory study conducted in June and July 2025 in Birjand, eastern Iran, near the Iran-Afghanistan border. Forty-two unstructured interviews were held with patients at a group of clinics and primary healthcare centers. The interviews were conducted during and immediately following the onset of the armed conflict between Iran and Israel. Data were analyzed using the Boyatzis framework.</p><p><strong>Results: </strong>The findings revealed that barriers to AI chatbot usage among patients could be categorized into four main areas: lack of awareness, knowledge, and training; user barriers; technological barriers; and financial barriers. User barriers encompassed the greatest number of sub-themes, whereas financial barriers were represented by only a single sub-theme, indicating less representation by the study participants.</p><p><strong>Conclusion: </strong>The study provided novel data on wartime conditions, offering valuable insights for stakeholders in conflict-affected or at-risk regions. The findings can support effective AI chatbot adoption by local populations during war or crises and inform efforts to improve usage in low-resource and conflict settings.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261485226"},"PeriodicalIF":3.9,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13527368/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148867939","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":"When documentation follows the patient: Unprofessional language and behavioral health referrals in substance use disorder - a retrospective cohort study using LLM-augmented natural language processing.","authors":"Jiyoun Song, Sue Hyon Kim, Yoonjae Lee, Mollie Hobensack, Hyunjin Song, Aviv Y Landau","doi":"10.1177/20552076261483333","DOIUrl":"10.1177/20552076261483333","url":null,"abstract":"<p><strong>Objective: </strong>Substance use disorders (SUDs) are a major public health challenge, and stigma remains a key barrier to care. Unprofessional or stigmatizing language can shape clinician perceptions and affect decision-making. Traditional natural language processing (NLP) often misses context-dependent bias, while large language models (LLMs) pose reliability concerns. This study aimed to (1) develop an LLM-enhanced, human-validated NLP model to detect unprofessional language, (2) quantify unprofessional language and behavioral health referrals, and (3) examine their association among patients with SUD.</p><p><strong>Methods: </strong>In this retrospective cohort study, we analyzed the MIMIC-IV, a large deidentified electronic health record database from a tertiary academic medical center in USA, for adult (≥18 years) with SUD admitted to the emergency department or intensive care unit between 2008 and 2019. A rule-based NLP algorithm detected unprofessional language. Three LLMs (GPT-4, Claude 3, Llama-3) expanded the vocabulary, and expert panel reviewed all generated terms for relevance and clinical realism before integration. Multivariable logistic regression examined the association between unprofessional language and behavioral health referrals.</p><p><strong>Results: </strong>Of 260,347 patients, 31.5% had SUD. Unprofessional language was more frequent in SUD notes (72% vs. 52%). The LLM-enhanced model improved over baseline (F-score 0.89 to 0.91). Within the SUD cohort, unprofessional language was significantly associated with higher odds of referral (adjusted odds ratio = 1.54) (all p < .001).</p><p><strong>Conclusion: </strong>Unprofessional language was common in SUD documentation and associated with behavioral health referrals. This human-validated, LLM-enhanced approach highlights how documentation-based stigma may influence care pathways and underscores the need for bias-aware, equitable communication strategies.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261483333"},"PeriodicalIF":3.9,"publicationDate":"2026-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13525294/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148857846","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":"Healthcare professionals' intention to adopt a smart diabetic foot screening system: A UTAUT-informed mixed-methods study.","authors":"Ting-Ting Yeh, Jawl-Shan Huang, Yun-Jui Wu","doi":"10.1177/20552076261479447","DOIUrl":"10.1177/20552076261479447","url":null,"abstract":"<p><strong>Background: </strong>Diabetic foot ulcers are serious but potentially preventable diabetes complications. Although smart screening technologies may support early detection, their routine clinical uptake remains limited, and factors shaping healthcare professionals' willingness to adopt such tools are insufficiently understood.</p><p><strong>Objective: </strong>This study examined determinants of healthcare professionals' behavioral intention to adopt FootHow, a smart diabetic foot screening system, using the Unified Theory of Acceptance and Use of Technology (UTAUT). Because participants evaluated FootHow after a standardized demonstration rather than hands-on clinical use, the outcome represents anticipated adoption intention, not observed adoption behavior.</p><p><strong>Methods: </strong>We conducted a cross-sectional survey of 80 healthcare professionals from two medical centers in Taiwan. Seven UTAUT-related constructs and behavioral intention were measured using a structured questionnaire. Quantitative data were analyzed using descriptive statistics, bivariate correlations, and multiple linear regression. Optional semi-structured interviews with a subset of participants were analyzed using inductive content analysis to contextualize the quantitative findings.</p><p><strong>Results: </strong>Participants reported generally favorable perceptions of FootHow. In bivariate analyses, attitude, effort expectancy, self-efficacy, and social influence were significantly associated with behavioral intention. In the multivariable model, attitude was the only independent predictor of behavioral intention (β=0.605, P<.01), and the model explained 36% of the variance in intention. Qualitative findings suggested that FootHow was perceived as valuable for early detection, patient awareness, and workflow efficiency, but concerns remained regarding accessibility for older adults, device reliability and hygiene, and affordability.</p><p><strong>Conclusions: </strong>Provider attitude was the strongest independent determinant of anticipated intention to adopt FootHow. Implementation strategies for preventive digital health technologies should therefore address provider attitudes while also strengthening workflow fit, usability, affordability, and transparent evidence regarding model performance and operational reliability. Longitudinal and pilot implementation studies are needed to determine whether these intentions translate into sustained clinical adoption.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261479447"},"PeriodicalIF":3.9,"publicationDate":"2026-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13525315/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148857851","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":"Short-term BMI changes associated with self-selected weight-control dietary patterns among Chinese office-based adults using a digital health platform: An exploratory real-world observational study.","authors":"Guyanan Li, Siqi Wang, Chunyan Liu, Rui Zhang, Haihua Ma, Yiwen Wang, Lixin Na","doi":"10.1177/20552076261484305","DOIUrl":"10.1177/20552076261484305","url":null,"abstract":"<p><strong>Background: </strong>Digital weight-management platforms increasingly provide multiple dietary strategies that users may select according to personal preference, cultural compatibility and perceived feasibility. However, real-world evidence regarding short-term weight changes associated with self-selected dietary patterns among Chinese office-based adults remains limited.</p><p><strong>Methods: </strong>This real-world observational study used data from an online weight-management platform in China between September and October 2024. Adults aged 18-60 years with self-reported office-based employment and available baseline and follow-up weight records were included. Participants voluntarily selected one of six dietary pattern groups: 16:8 time-restricted eating, 5:2 intermittent fasting, flexitarian diet, moderate-to-high protein diet, calorie-restricted balanced diet (CRD), or Dietary Approaches to Stop Hypertension (DASH). Within-group BMI changes were tested using paired t-tests, and between-group BMI changes were assessed using analysis of covariance (ANCOVA) adjusting for age, sex, baseline BMI, and follow-up duration.</p><p><strong>Results: </strong>A total of 5066 participants were included (mean age 31.53± 10.69 years; 94.20% female), and the CRD (39.42%) were the most commonly selected dietary patterns, whereas the flexitarian and DASH groups were substantially smaller. The median follow-up duration was 19 days. Significant within-group reductions in BMI were observed in the 16:8 time-restricted eating (-0.46 kg/m²), moderate-to-high protein (-0.28 kg/m²), CRD (-0.30 kg/m²), and flexitarian (-1.06 kg/m²) groups. After adjustment for age, sex, baseline BMI, and follow-up duration, BMI change differed across dietary pattern groups (ANCOVA: <i>F</i> = 3.648, <i>P</i> = 0.003).</p><p><strong>Conclusions: </strong>In this exploratory real-world observational study, several self-selected dietary patterns were associated with modest short-term reductions in BMI among Chinese office-based adults using a digital weight-management platform. Because dietary pattern selection was voluntary and follow-up was short, the findings should be interpreted as observational associations rather than evidence of comparative dietary effectiveness. Further studies with longer follow-up and more rigorous comparative designs are needed.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261484305"},"PeriodicalIF":3.9,"publicationDate":"2026-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13525286/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148857915","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":"Trends and hotspots in artificial intelligence applications for atherosclerosis research: A bibliometric analysis.","authors":"Yehan Lv, Siyuan Sun, Yuzhuo Zhang, Qun Niu, Jianfeng Li, Liqun Zhong","doi":"10.1177/20552076261483899","DOIUrl":"10.1177/20552076261483899","url":null,"abstract":"<p><strong>Background: </strong>Atherosclerosis (AS) is a complex systemic, immune-inflammatory vascular disease, and there is an urgent need to innovate its diagnostic and therapeutic strategies. The rapid advancement of artificial intelligence (AI) technology has opened up new avenues for the early diagnosis, risk prediction, and precision treatment of AS. However, a systematic quantitative analysis of global knowledge and collaboration in this field is still lacking.</p><p><strong>Methods: </strong>This study conducted a systematic search of the Web of Science Core Collection (WoSCC), Scopus, and PubMed databases for literature on the application of AI in AS research published between 2000 and 2025. Advanced bibliometric tools, such as CiteSpace, VOSviewer, and R-Bibliometrix, were employed to conduct multidimensional visual analyses of publication trends, transnational collaboration networks, knowledge flows in core journals, and the emergence of keywords.</p><p><strong>Results: </strong>A total of 258 core publications were ultimately included, involving 1,900 authors, 44 countries/regions, and 1,152 research institutions. The spatio-temporal distribution revealed that the volume of publications in this field has grown significantly, with rapid growth beginning after 2021. Globally, a China-U.S. dual-center pattern has emerged: China leads in terms of output scale and growth rate, while the United States holds higher citation accumulation in terms of total citations and average citations per paper. However, raw citation counts and average citations per paper are influenced by publication year, document type, and denominator size; these cross-country differences should therefore be interpreted with caution, and future studies using normalized indicators are warranted. The evolution of research hotspots has undergone a distinct three-stage transition: from early basic algorithm development (2000-2014), through multimodal image intelligent analysis (2015-2020), to the current phase of clinical multicenter validation and in-depth mechanism decoding (2021-2025). Journal overlay analysis further confirms that clinical medicine is increasingly intersecting with basic life sciences, driving bidirectional knowledge transfer.</p><p><strong>Conclusions: </strong>This study systematically maps the global research landscape of AI in the field of AS, revealing its evolutionary path from methodological exploration to clinical translation. Future research should break down data barriers, advance multinational, multiethnic, and multicenter cohort validation, and focus on the development of interpretable AI models. The observed trends in the bibliometric data suggest that the integration of AI technology with systems biology and a holistic medical approach may play an increasingly important role in personalized, precision interventions for AS.</p>","PeriodicalId":51333,"journal":{"name":"DIGITAL HEALTH","volume":"12 ","pages":"20552076261483899"},"PeriodicalIF":3.9,"publicationDate":"2026-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13525297/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148857900","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}