Frontiers in digital health最新文献

Background: Prior authorization (PA) is a major source of administrative burden, treatment delay, and clinician burnout. Artificial intelligence (AI), particularly large language models (LLMs), is increasingly used to assist with clinical documentation, yet its reliability for payer-facing administrative tasks remains uncertain.

Objective: To evaluate the quality of PA letters drafted by ChatGPT-5 for commonly used medications requiring PA in nephrology. Quality was evaluated based on correctness and strength of clinical reasoning.

Methods: We created a single standardized prompt and applied it across 29 nephrology scenarios to generate PA letters. Each PA letter was reviewed against four criteria: 1) absence of false statements or hallucinations, 2) correctness of ICD-10 coding, 3) presence and validity of citations, and 4) clinical reasoning, rated on a 4-point Likert scale (illogical, weak, adequate and strong). FDA drug labels, KDIGO guidelines and related randomized controlled trials were used as reference standards.

Results: Out of 29 letters, one letter (3.5%) contained false statements mentioning an irrelevant clinical trial. The ICD-10 diagnosis code was correct in 23 letters (79.3%), most errors were related to chronic kidney disease (CKD) staging or internal diagnostic inconsistencies. 27 letters (93.1%) cited valid references, with one letter citing an incorrect trial and another one citing a correct KDIGO guideline with inaccessible link. Twenty-six letters (89.7%) demonstrated strong clinical reasoning, supported by guideline-oriented or FDA label-aligned justification. The remaining 3 letters were rated as adequate reasoning. The main areas for improvement involved citing relevant references and emphasizing special considerations, for example Risk Evaluation and Mitigation Strategy (REMS) compliance for eculizumab.

Conclusions: ChatGPT-5 can generate clinically coherent PA drafts for nephrology medications, but limitations in coding precision and citation reliability persist. With appropriate oversight, AI-assisted documentation may reduce administrative burden while maintaining safety and accuracy.

While most existing digital immunization systems lack mechanisms to capture high fidelity real-time data to respond to current needs; many others are not designed to support interoperability and data sharing across the continuum of care for the health Sustainable Development Goals (SDGs). In this paper, we used the World Health Organization (WHO) Digital Documentation of COVID-19 Certificates (DDCC) as a proxy to demonstrate and operationalize how an efficient digital immunization system could strengthen service delivery and optimize outcomes for the SDGs. This paper appraises the technical, ethical and cultural considerations for establishing DDCC and how it can be operationalized among national health systems. It demonstrates how digital health investments can support routine immunization for the SDGs and highlights the critical role global health leadership plays in shaping reforms for national digital transformation agenda. The adoption and institutionalization of digital immunization systems offer opportunities to bridge multiple information solutions and strengthen immunization service delivery towards sustainable outcomes for the SDGs. Thus, it is recommended that Development partners and implementers jointly work with governments to shape the national digital health ecosystem that connects multiple healthcare journeys for the sustainable immunization agenda 2030.

[This corrects the article DOI: 10.3389/fdgth.2025.1653168.].

Background: Short videos have become a primary channel for Lumbar Disc Herniation (LDH) patients to obtain disease knowledge and rehabilitation guidance. Algorithm-driven personalized recommendations may expose patients to heterogeneous LDH-related content, affecting their health anxiety and rehabilitation trajectories.

Objective: This study explored the impacts of LDH-related short video exposure duration and content types on health anxiety and lumbar functional rehabilitation in LDH patients, and verified the mediating role of health anxiety.

Methods: A 6-month prospective cohort study enrolled 213 LDH outpatients from Luoyang Orthopedic-Traumatological Hospital (Jan-Apr 2025). Demographic, clinical and short video usage data were collected. Health anxiety (MCQ-HA) and lumbar function (JOA) were assessed at baseline and follow-up. Pearson correlation, multiple linear regression, subgroup analysis and Bootstrap mediation analysis (5,000 resamplings) were used.

Results: At 6-month follow-up, the mean JOA score decreased from 23.00 ± 1.59 at baseline to 21.96 ± 3.03, and the mean MCQ-HA score increased from 20.77 ± 4.57-21.86 ± 6.14. Pearson correlation analysis showed that daily viewing duration and exposure frequency to awareness-motivation content were significantly negatively correlated with ΔJOA (r = -0.36, r = -0.33; both P < 0.001) and positively correlated with ΔMCQ-HA (r = 0.31, r = 0.34; both P < 0.001). Multiple linear regression indicated that ΔJOA in the >60 min daily viewing group was significantly lower than that in the <30 min group; exposure frequency to awareness-motivation content was independently negatively associated with ΔJOA and positively associated with ΔMCQ-HA (both P < 0.001), with no significant associations found for other content categories (all P > 0.05). Subgroup analysis based on clinical efficacy criteria revealed significant differences in recovery outcomes across viewing duration groups (χ ² = 18.75, P = 0.004). Bootstrap mediation analysis confirmed that ΔMCQ-HA mediated 16.13% of the total effect of daily viewing duration on ΔJOA and 20.80% of the total effect of awareness-motivation content exposure frequency on ΔJOA.

Conclusion: Prolonged short video exposure and frequent awareness-motivation content viewing were associated with poorer rehabilitation and higher health anxiety, with health anxiety partially mediating these relationships, providing empirical evidence for digital health guidance.

Purpose: The Resilient Minds (ReMind) cohort was established to investigate cognitive and mental health resilience across the life course, addressing a gap in longitudinal evidence about resilience. The study collected data on traditional medical and lifestyle risk factors for chronic disease, genetics, and a range of mental health and cognitive outcomes. It also aimed to explore contemporary contextual influences on resilience, including internet use, social engagement, environmental exposures, and life course adversities such as perceived discrimination.

Participants: The cohort included 1,640 adults aged 18-93 years, recruited through social media and community groups, to participate in a fully remote, two-year health study. Participants completed online surveys, cognitive and sensory testing, and intensive "sprints" occurring approximately every three months, during which daily surveys and digital health data were collected. A brain-health substudy (BHS) is being conducted for participants aged 50 years and older (current n = 184/400 planned), involving to evaluate neuroimaging, blood and ocular biomarkers to assess resilience and cognitive decline.

Findings to date: Thirty percent of participants were born overseas, and the average years of education were 14.7, 15.0 and 14.1 for young, middle aged and older adults, respectively. Among adults aged 65 years and older, 41.9% reported hypertension, 39.1% high cholesterol, 7.1% diabetes, and 22.4% obesity. In the BHS, 18% met criteria for Subjective Cognitive Decline, and 15% met criteria for Mild Cognitive Impairment.

Future plans: The initial study duration is 2 years, with plans to seek funding for extended follow-up to identify long-term predictors of cognitive and mental health resilience and the development of cognitive impairment in ageing.

This article provides a critical narrative synthesis of literature on healthcare management in Africa, focusing on two interconnected areas: the impact of managerial capability on shaping integrated healthcare ecosystems and the adoption, implementation, and governance of digital health innovations within diverse health system contexts. Based on health systems strengthening frameworks and socio-technical views on digital transformation, the article explores how managerial skills influence the development and outcomes of digital health initiatives across African settings. Rather than presenting new empirical data, it uses comparative analysis of existing studies to highlight opportunities and ongoing challenges, such as uneven managerial digital skills, resistance to change, system fragmentation, and unintended effects like digital exclusion. The article concludes with a clear and practical agenda for future research and policy, emphasising the vital role of digitally competent managers in fostering supportive organisational cultures, promoting system integration, and ensuring meaningful adoption of digital health innovations by frontline health workers and patient populations.

Background: Medical text summarization using large language models (LLMs) has reached an inflection point in 2024-2025, with adapted models demonstrating capability to match or exceed human expert performance in specific tasks. However, critical gaps persist in safety validation, evaluation frameworks, and clinical deployment readiness. A comprehensive review revealed that only 7% of studies conducted external validation and 3% performed patient safety assessments, with hallucination rates ranging from 1.47% to 61.6%. Existing reporting guidelines, including CONSORT-AI, SPIRIT-AI, TRIPOD-LLM, and DEAL, do not adequately address the specific requirements of medical text summarization tasks.

Objective: to develop MEDAI-LLM-SUMM, the first specialized reporting checklist for research on medical text summarization using LLMs, addressing critical gaps in existing reporting standards.

Methods: A modified iterative consensus approach was employed, comprising three sequential stages: (1) a systematic literature review of 216 publications from PubMed and eLibrary (2023-2025) following PRISMA guidelines and an analysis of existing reporting standards (TRIPOD-LLM, DEAL, CONSORT-AI, SPIRIT-AI, TRIPOD + AI, CLAIM, STARD-AI); (2) development of an initial 44-item, 7-section checklist by a supervisory group; (3) three rounds of face-to-face consensus discussions with a multidisciplinary expert panel of 11 specialists (3 radiologists, 2 clinicians, 3 medical informatics experts, 1 biostatistician, and 2 medical LLM developers). The consensus criterion required unanimous agreement from all panel members.

Results: The final MEDAI-LLM-SUMM checklist comprises 24 items organized into six sections: (A) Clinical validity (4 items addressing clinical task definition, expert involvement, hypothesis formulation, and medical expertise requirements); (B) Model Selection (5 items covering model justification, system requirements, deployment environment, LLM-as-judge approach, and prompt documentation); (C) Data (3 items on datasets, reference summaries with expert consensus, and data stratification); (D) Quality Assessment (8 items including evaluation metrics, clinical metrics, expert evaluation, hallucination detection, LLM-judge assessment, sample size justification, pilot testing, and limitations documentation); (E) Safety (2 items on ethical approval and data anonymization); and (F) Data Availability (2 items on code and dataset accessibility). Comparative analysis with six existing reporting standards demonstrated that MEDAI-LLM-SUMM uniquely addresses hallucination assessment requirements, reference summary creation methodology, LLM-as-judge validation protocols, and detailed pilot testing specifications.

Background: Deep learning (DL) methods for surgical video analysis have expanded rapidly in minimally invasive surgery (MIS). However, a structured bibliometric overview focused on DL-based surgical instrument segmentation, detection, and tracking is lacking. The objective of this review is to systematically map the research landscape with this focus, by examining publication trends, influential authors, institutions, and countries, collaboration networks, keyword co-occurrence patterns, and the thematic trajectory of the discipline.

Methods: We performed a bibliometric analysis of original research articles on DL-based surgical instrument segmentation/detection/tracking in laparoscopic or robotic MIS, published between 2017 and 2024. Searches were conducted in six databases namely PubMed, Scopus, IEEE Xplore, Embase, Medline, and Web of Science. Records were de-duplicated in EndNote and analyzed using the Bibliometrix R package, with co-authorship, co-citation, and keyword networks visualized in VOSviewer. Citation counts were extracted from each study's respective database and interpreted cautiously given the influence of publication age.

Results: We included 217 articles. Annual output increased from 2017 to a peak in 2023, indicating sustained growth in DL research for MIS instrument analysis. The most productive countries included the United States and France, with major institutional contributions from the University of Strasbourg and Furtwangen University. Keyword analysis indicated continued dominance of convolutional neural networks alongside emerging themes including transformer-based architectures, multimodal learning, and real-time intraoperative applications.

Conclusions: This bibliometric study characterizes the evolution, leading contributors, collaboration patterns, and thematic trajectories of DL-based instrument segmentation/detection/tracking in MIS. While these findings can inform research prioritization and collaboration, this study does not evaluate clinical effectiveness. Future work should prioritize explainable and efficient real-time models, standardized annotation protocols, and broader global partnerships to support responsible clinical translation.

Parkinson's disease (PD) is a progressive neurodegenerative disorder characterized by motor and non-motor impairments, where early diagnosis remains challenging due to reliance on subjective clinical assessments. Recent artificial intelligence (AI)-based approaches have demonstrated promise in identifying subtle PD biomarkers from individual modalities such as speech, gait, and handwriting; however, unimodal systems often fail to capture the heterogeneity of the disease and provide limited interpretability. To address these limitations, this study proposes a multimodal deep learning framework that integrates handwriting, gait, and speech modalities using an early feature fusion strategy for robust and interpretable PD detection. Each modality is processed through a dedicated feature extraction pipeline using deep neural networks, followed by static feature concatenation and classification using an XGBoost model. Model transparency is enhanced using explainable AI (XAI) techniques, including SHapley Additive exPlanations (SHAP) and Gradient-weighted Class Activation Mapping (Grad-CAM), enabling clinical interpretability of modality- and feature-level contributions. Experimental evaluation on benchmark datasets demonstrates that the proposed trimodal fusion model achieves an accuracy of 92%, outperforming unimodal handwriting (91%), gait (90%), and speech (74%) models. The fusion framework attains a macro F1-score of 0.89, an area under the ROC curve (AUC) of 0.95, and an average precision (AP) of 0.96, indicating strong discriminative capability and robustness. Confusion matrix analysis reveals balanced sensitivity (90%) and specificity (89%) across classes. Explainability analysis confirms that handwriting tremor patterns, gait force asymmetries, and speech spectral instabilities are key contributors to PD prediction. These results highlight the effectiveness of explainable multimodal AI in delivering accurate, reliable, and clinically interpretable solutions for early PD detection.

The advance of medicine and technology has been a catalyst for the population's longevity, reflected in the increase in elderly citizens. However, this increase also comes a burden on caregivers. To address this gap between need and accessibility, the motivation for this study arises, highlighting the challenges faced by an ageing population. This work presents a preliminary proof of concept of an innovative digital tool (a mobile app prototype) to support older people to live more independently and safely, while facilitating communication between them, caregivers and health professionals. To develop the prototype for monitoring elderly health, the User-Centred Design methodology was applied, concluding with a usability evaluation. As a proof-of-concept, this study suggests that combining technology with human support may contribute to improved elderly care and empowerment; however, these implications remain preliminary and require validation in larger and more diverse evaluations. Theoretically, it uses a social determinant of health lens to outline potential ways in which health apps could support access to care in this age group, to be examined in future, larger-scale evaluations. From a practical perspective, it contributes with a preliminary proof of concept and prototype offering use-cases aimed at improving the quality of life of this population.