Studies in health technology and informatics最新文献

Unplanned interruptions in radiation therapy (RT) increase clinical risks, yet proactive, personalized psychosocial support remains limited. This study presents a proof-of-concept framework that simulates and evaluates Empathic AI-patient interactions using large language models (LLMs) and synthetic oncology patient personas. Leveraging a de-identified dataset of patient demographics, clinical features, and social determinants of health (SDoH), we created realistic personas that interact with an empathic AI assistant in simulated dialogues. The system uses dual LLMs, one for persona generation and another for empathic response, which engage in multi-turn dialogue pairs per persona. We evaluated the outputs using statistical similarity tests, quantitative metrics (BERTScore, SDoH relevance, empathy, persona distinctness), and qualitative human assessment. The results demonstrate the feasibility of scalable, secure, and context-aware dialogue for early-stage AI development. This HIPAA/GDPR compliant framework supports ethical testing of empathic clinical support tools and lays the groundwork for AI-driven interventions to improve RT adherence.

Large Language Models (LLMs) are promoted as solutions to many problems in medicine and wider health care. However, the empirical evidence of these claims is currently limited, as clinical trials usually take several years until publication. Clinical trial registries, such as ClinicalTrials.gov, allow for a glimpse into the topics on which publications can be expected in the future. The aim of the present study is to identify studies on ClinicalTrials.gov that use LLMs and to summarize their characteristics and topics. We identified 94 studies involving LLMs after keyword-based screening and subsequent manual inspection. All studies had start dates in 2023 or later. Compared to other studies, LLM-studies relatively often had the primary purpose "health services research", while "treatment" was relatively rare. The most common topics of LLM-studies were diagnostics, clinical recommendations, and other supportive functions. These findings underscore that LLMs are currently not being evaluated for treatment, prevention, or drug discovery, but rather for their linguistic and reasoning capabilities as assistive tools.

The National HTA Programme (PNHTA) - Medical Devices is designed to promote collaboration among the entities responsible for decision-making processes, with the purpose of developing and implementing tools based on Health Technology Assessment (HTA), ensuring more effective governance of medical devices. This study focuses on the implementation of a new strategy for managing the procurement requests of innovative medical devices, in line with the PNHTA. Specifically, it aims to develop a support method for healthcare organizations planning to introduce new technologies into clinical practice, providing a useful tool to guide decisions regarding the adoption or exclusion of each device. The innovation lies in identifying a method aimed at improving the robustness of healthcare decisions. The proposed model uses the Analytic Hierarchy Process (AHP) method to conduct a multicriteria analysis of the innovative devices, in order to strengthen the decision-making process. This method allows for the comparison and evaluation of different alternatives based on specific criteria and sub-criteria, with the objective of identifying the most advantageous solution.

Social media platforms such as TikTok are increasingly used to access health information, particularly among younger and digitally connected populations. However, the unregulated nature of this content raises concerns about medical misinformation. This study applied an AI-assisted framework to evaluate the clinical accuracy of 619 TikTok transcripts related to diabetic foot care, using authoritative guidelines from the ADA, IWGDF, and IDSA. Findings show that while some videos convey partially accurate information, over 42% contained misleading or false claims, including advice that could delay treatment or worsen outcomes. Semantic analysis highlighted a prevailing focus on complications and amputation, with minimal attention given to preventive care and early intervention. These results underline the pressing need to address misinformation and promote responsible digital health education.

The convergence of mobile electroencephalography (EEG) technology and artificial intelligence (AI) offers transformative potential for education. We propose a novel conceptual framework that integrates DreamMachine, a clinically validated mobile EEG device, with AI-driven adaptive learning systems. Our vision is to create neuroadaptive educational environments where real-time EEG signals, including markers of attention, cognitive load, and emotional states, and mental well-being inform AI algorithms to personalize content delivery dynamically. Such an approach could significantly enhance learning efficiency, engagement, and inclusivity, and support the mental health of learners by identifying stress or cognitive overload early and enabling timely, personalized interventions. This study outlines the technical feasibility of leveraging DreamMachine's high-fidelity, low-cost, portable EEG data in the classroom and remote settings. It proposes a machine-learning pipeline for real-time cognitive state detection. Ethical considerations surrounding neurodata use in education are discussed, emphasizing the need for privacy, transparency, and student agency. We invite collaboration on this interdisciplinary initiative, aiming to pilot the system in educational settings and redefine the future of personalized, mentally supportive learning.

Clinical research often requires integrating data from diverse sources, which differ not only in structure but also in semantics and language. Traditional extract-transform-load (ETL) pipelines struggle to handle semantic variability and lack built-in support for multilingual or ontology-driven harmonisation. This fragmentation limits the interoperability and reuse of clinical datasets in large-scale analyses. In this paper, we propose an integrated framework that combines an embedding-based concept mapping engine with an automated ETL pipeline using Apache Airflow. The mapping engine uses transformer-based embeddings to align clinical terms with standard concepts, producing outputs in White Rabbit and Usagi-compatible formats to ensure backward interoperability. We validated the system using multilingual real-world datasets demonstrating its ability to handle heterogeneous inputs and maintain end-to-end reproducibility.

Open health data is arguably the cornerstone of modern public health strategy, enabling data-driven policymaking and promoting transparency. The extent and format of health data publication, however, vary widely across jurisdictions, especially within multi-tiered health governance systems. This study subsequently investigates the digital presence and open data strategies of health authorities in Germany and Switzerland, focusing on their official websites as the primary interface for public communication. A structured content assessment was conducted for 16 German state-level public health authorities (Landesgesundheitsämter), 26 Swiss cantonal health departments, and both countries' national public health bodies. Findings show that having a web presence, health-related content, data dashboards, and access to raw (machine-readable) datasets are prominent. German state-level authorities frequently publish general health information with statistical reports and datasets, while Swiss cantons largely offer general health information. At national-level, however, Switzerland provides centralized open data access, unlike Germany's more distributed model (i.e. at state-level). The results suggest that open data visibility is strongly influenced by the structure of public health governance based on population size - decentralized in Germany and more centralized in Switzerland. These findings highlight the value of observing communication trends across governance tiers (and population sizes) to inform open health data strategies in federated systems, and beyond.

The adoption of digital technologies in healthcare is growing rapidly, and with it, the associated cybersecurity risks are also increasing. In particular, web applications, which can be used to manage and share sensitive health and personal information, require strong security measures to prevent data breaches and ensure compliance with regulatory standards. This paper investigates the applicability of the Open Web Application Security Project (OWASP) guidelines in the healthcare domain. Through a literature review, we identified the most common security requirements considered and used in Digital Health (DH) technologies and assessed their alignment with OWASP Application Security Verification Standard (ASVS). Furthermore, a questionnaire, involving Italian healthcare facilities and Information Technology (IT) companies operating in the healthcare sector, highlighted a significant gap between the availability of security standards and guidelines, and their actual knowledge and use in practice. Based on these findings, we propose a context-aware tool that guides developers and testers in applying OWASP standards throughout the software development lifecycle. The proposed tool aims to provide tailored security recommendations, structured checklists, and test planning based on application context, offering a practical bridge between frameworks and real-world adoption in clinical environments.

Diagnosing headache disorders remains a clinical challenge due to the heterogeneity of headache phenotypes and the absence of objective biomarkers. This study presents a curated dataset of 50 clinical headache case examples, comprising both real (n = 34) and synthetic (n = 16) cases, categorized across 20 diagnoses according to ICHD-3 criteria. The dataset enables the evaluation of large language models (LLMs) for diagnostic accuracy in headache medicine. Three GPT-based models were tested using different prompting strategies, with diagnostic performance assessed at both diagnosis and group levels. Top-1 accuracy ranged from 24% to 63% at the diagnosis level and up to 92% at the group level. The results highlight the potential of LLMs in supporting differential diagnosis of headache disorders, while also emphasizing the need for further validation with larger, diverse datasets. Future efforts will focus on expanding real-world data through clinical collaborations and benchmarking LLMs against medical professionals to assess their utility in clinical decision-making.

The digitization of healthcare - through electronic health records, predictive algorithms, remote monitoring, and automated decision-making tools - has revolutionized clinical workflows and optimized patient management. However, these developments often carry unintended consequences when applied to the end-of-life context, where the subjective, relational, and existential dimensions of dying resist abstraction and quantification. This paper explores the tensions between digital efficiency and the human realities of death, arguing that the virtuality of digital health systems risks alienating patients, families, and clinicians at precisely the moments where care must be most embodied and relational. Drawing from a conceptual analysis informed by medical ethics and palliative care literature, we examine how virtual representations (data, dashboards, protocols) interact with real dying bodies and social relationships. Through case illustrations, we highlight how systems designed for efficiency can unintentionally marginalize suffering, flatten complex narratives, and displace the rituals and presence that define authentic death. Our findings suggest a pressing need to reorient digital health design to account for the limits of representation and the irreplaceability of human connection at the end of life. We argue that any future model of digital care must not only prioritize outcomes but also preserve dignity, ambiguity, and relational integrity in death.