focalinterest.net

Tag: Health Services & Management

  • Patients supported AI-drafted portal messages with clinician oversight

    What the study found

    Patients viewed patient portal messaging as mainly transactional, with a focus on timely problem-solving rather than relationship-building. They were generally comfortable with AI-drafted replies, but that comfort depended on clinician oversight and accountability.

    Why the authors say this matters

    The authors conclude that AI use in portal messaging may support more efficient communication if clinicians remain responsible for the messages. They also say implementation should include clinician oversight, context-sensitive communication standards, standardized disclosure practices, and monitoring of quality, equity, patient trust, and safety.

    What the researchers tested

    The study used a qualitative design with vignette-based prompts modeled on scenarios from a prior survey about AI-drafted online messaging. Researchers interviewed 40 adult patients from a large academic health system by videoconference between April and August 2025, using draft messages that varied in tone, length, and implied authorship.

    What worked and what didn't

    Patients expressed high comfort with AI-drafted messages when a clinician reviewed them. Preferences for tone and empathy varied by person and situation, and were shaped more by whether the message fit the purpose and seriousness of the communication than by whether it seemed AI-like or human-like. Participants broadly supported AI disclosure, but they differed on when and how disclosure should be presented.

    What to keep in mind

    This was a qualitative study of 40 patients from one large academic health system, so the findings describe this sample rather than a broader population. The abstract does not describe long-term outcomes, and it notes that downstream effects on quality, equity, trust, and safety should be monitored.

    • Patients described portal messages as transactional and focused on timely problem resolution.
    • Comfort with AI-drafted replies was high when clinicians reviewed the messages.
    • Message tone, length, and detail mattered more than whether a message seemed AI-like or human-like.
    • Participants broadly wanted disclosure that AI was involved, but preferred different timing and formats.
    • The authors recommend clinician oversight and monitoring of effects on quality, equity, trust, and safety.

  • TAIGHA measures trust in AI-generated health advice

    What the study found

    The study found that the Trust in AI-Generated Health Advice scale (TAIGHA) and its four-item short form (TAIGHA-S) are validated questionnaires for measuring users' state trust and distrust in AI-generated health advice. The full scale and short form showed strong psychometric properties.

    Why the authors say this matters

    The authors say this matters because people increasingly use AI tools, including large language models, to get health information and support health-related decisions. The study suggests that measuring trust in AI-generated health advice is important because advice-taking can have clinical, safety, and healthcare-system consequences.

    What the researchers tested

    The researchers developed theory-based questionnaire items using a generative AI approach and then validated them in several steps. These included automated validation, content validation with 10 domain experts, face validation with 30 lay participants, and psychometric validation with 385 UK participants who received AI-generated health advice for symptom assessment.

    What worked and what didn't

    After automated item reduction, 28 items were retained and then reduced to 10 based on expert ratings. The final TAIGHA scale showed excellent content validity, face validity, and model fit, and it had high internal consistency for both trust and distrust; the short form correlated highly with the full scale and also showed high reliability. The abstract does not describe any major elements that did not work, beyond the item-reduction process that narrowed the scale.

    What to keep in mind

    The psychometric validation was conducted with 385 participants in the U.K. receiving AI-generated health advice for symptom assessment, so the reported validation is specific to that sample and context. The abstract does not describe other limitations.

    • TAIGHA and TAIGHA-S were developed to measure trust and distrust in AI-generated health advice.
    • The study used automated validation plus expert, lay, and participant testing.
    • The final TAIGHA scale showed excellent content validity, face validity, and model fit.
    • Both trust and distrust subscales showed high internal consistency.
    • The short form correlated strongly with the full scale and was also reliable.

  • AI may support public health only with strong governance

    What the study found

    The article argues that artificial intelligence (AI) can become a foundational shift for public health only if it is explicitly aligned with public health values such as prevention, equity, transparency, and accountability. If used uncritically, the authors say it risks becoming a technocratic distraction that emphasizes data-driven efficiency over social context.

    Why the authors say this matters

    The authors conclude that public health decisions affect whole populations and require ethical judgment, political legitimacy, and community trust. They suggest that AI should support, rather than weaken, these commitments by staying tied to democratic deliberation and social accountability.

    What the researchers tested

    This is an opinion article, not an empirical study. The author examines methodological tensions, normative conflicts, and governance challenges around AI in public health, drawing on existing scholarship and examples such as machine learning, large multimodal models (AI systems that combine multiple kinds of data), and precision public health.

    What worked and what didn't

    The article says AI can be useful for pattern recognition, speed, and scale, especially in real-time disease monitoring, outbreak forecasting, and system optimization. It also notes that AI is more aligned with public health when used for collective risk assessment and structural intervention, rather than only individual-level risk prediction. At the same time, the authors highlight problems with black-box models, biased training data, weak generalizability, high infrastructure costs, and the lack of standardized validation and oversight.

    What to keep in mind

    The available text does not describe new original data or a formal evaluation of a specific AI system. The article’s claims are conceptual and based on cited literature, so its limitations are those of an opinion piece rather than a measured intervention study.

    • The article says AI could be a foundational shift for public health only if it fits public health principles.
    • It warns that uncritical AI adoption may privilege efficiency over social context and equity.
    • The authors emphasize transparency, explainability, and human oversight as necessary for legitimacy.
    • The article notes that biased or incomplete data can reproduce or worsen health inequities.
    • It highlights benefits for outbreak monitoring and other population-level uses, but also limits from opacity, data quality, and infrastructure needs.

  • Accreditation standards align with AI readiness in hospitals

    What the study found

    The study found conceptual overlaps between Brazilian hospital accreditation requirements and the organizational enablers needed for AI adoption. The authors argue that accreditation can help build readiness for AI through governance, information systems, process improvement, and staff development.

    Why the authors say this matters

    The authors say this matters because accreditation may provide the governance and process foundations needed for effective AI use in healthcare. They also suggest that these overlaps may support a hospital's broader digital transformation journey.

    What the researchers tested

    The researchers used an interpretive analysis to compare the 2022-2026 Brazilian National Accreditation Organization (ONA) Accreditation Manual with a 2024 systematic review by Rahimi et al. on enablers and barriers to AI implementation in hospitals. They grouped the AI factors into People, Process, Information, and Technology dimensions and mapped them against ONA standards related to leadership, quality and safety, information security, technology management, and workforce development.

    What worked and what didn't

    The analysis identified alignments in several areas: leadership and formal planning, data collection and information security, technology acquisition and maintenance, continuous quality improvement, risk management, and staff training. The paper also notes that accreditation is not a direct roadmap for AI, does not guarantee AI success, and does not replace the need for AI-specific governance, validation, and ethical frameworks.

    What to keep in mind

    This is an interpretive and exploratory analysis, so it shows conceptual correspondences rather than direct evidence of AI outcomes. The authors caution that the strength of the alignment may vary by national context and that the findings are based on the Brazilian accreditation framework, though they suggest some mechanisms may transfer to other accreditation models.

    • The study maps Brazilian hospital accreditation requirements to organizational enablers for AI implementation.
    • The authors identify alignments in governance, data infrastructure, process improvement, and workforce readiness.
    • ONA standards on leadership, documentation, and planning are linked to AI governance needs.
    • ONA standards on information security and data governance are linked to data quality, interoperability, and privacy concerns.
    • The authors caution that accreditation supports readiness but does not replace AI-specific strategy or oversight.
  • AIoT smart elderly care in Beijing faces multiple implementation barriers

    AIoT smart elderly care in Beijing faces multiple implementation barriers

    What the study found

    The study found five major barriers to implementing artificial intelligence of things (AIoT, which means connected artificial intelligence systems) in elderly care in Beijing’s Xicheng District. These barriers were systemic fragmentation and a lack of unified standards, a mismatch between services and older adults’ needs, low adoption due to digital literacy and trust issues, a shortage of interdisciplinary caregivers, and heavy reliance on government subsidies with limited market participation.

    Why the authors say this matters

    The authors conclude that coordinated action across multiple stakeholders is needed to use AIoT to build sustainable, responsive, and human-centered elderly care ecosystems. They also propose that standardizing systems, improving digital inclusion, and strengthening caregiving capacity could help support this goal.

    What the researchers tested

    The researchers used a qualitative study design grounded in socio-technical systems theory, the Unified Theory of Acceptance and Use of Technology (a model of technology adoption), and welfare pluralism theory. They conducted semi-structured interviews and field observations in Beijing’s Xicheng District, which they describe as an urban area with a high aging rate and active smart elderly care initiatives.

    What worked and what didn't

    The study reports that AIoT-related elderly care efforts were constrained by five recurring problems rather than by a single issue. The proposed policy framework emphasizes standardizing AIoT systems, tailoring services to older adults’ needs, increasing financial investment, empowering communities, improving digital inclusion, and professionalizing the caregiving workforce.

    What to keep in mind

    The abstract describes a qualitative study in one district of Beijing, so the findings are limited to that setting in the available summary. It does not provide outcome measurements or indicate whether the proposed policy framework was tested.

    • The study identified five major barriers to AIoT-based smart elderly care in Beijing’s Xicheng District.
    • Older adults’ low technology adoption was linked to digital literacy and trust issues.
    • The researchers reported a mismatch between elderly care services and older adults’ needs.
    • There was a shortage of interdisciplinary professional caregivers.
    • The authors proposed a policy framework focused on standardization, digital inclusion, and workforce professionalization.

  • Burnout among Norwegian GPs rose from 2012 to 2024

    What the study found

    The study found that burnout among general practitioners in Norway increased substantially between 2012 and 2024. In 2024, burnout was linked with low job satisfaction, high work-related stress, and frequent sickness presenteeism, meaning working while sick.

    Why the authors say this matters

    The authors conclude that addressing modifiable factors such as work-related stress, job satisfaction, and sickness presenteeism is essential for sustaining physician well-being and maintaining patient care quality. They also note that burnout has implications for healthcare system sustainability.

    What the researchers tested

    The researchers used data from the Norwegian Physician Panel, a nationally representative survey from 2012, 2018, and 2024, and included only respondents who identified as general practitioners. Burnout was measured with the Maslach Burnout Index, and logistic regression was used in 2024 to examine associations with age, sex, weekly work hours, self-rated health, sick leave, presenteeism, job satisfaction, and work-related stress.

    What worked and what didn't

    Overall burnout rose from 5.8% in 2012 to 17.1% in 2018 and 21.8% in 2024. High emotional exhaustion increased from 19.1% to 47.2%, high depersonalisation from 2% to 24%, and low personal accomplishment became less common, falling from 16.4% to 6.3%; in 2024, burnout was significantly associated with low job satisfaction, high work-related stress, and frequent sickness presenteeism.

    What to keep in mind

    The abstract reports associations for 2024 but does not describe causal effects. It also does not provide detailed results for every tested factor, and no specific limitations are described in the available summary.

    • Burnout among Norwegian general practitioners increased from 5.8% in 2012 to 21.8% in 2024.
    • High emotional exhaustion rose from 19.1% to 47.2% over the same period.
    • High depersonalisation increased from 2% to 24%, while low personal accomplishment became less common.
    • In 2024, burnout was significantly associated with low job satisfaction, high work-related stress, and frequent sickness presenteeism.
    • The study used nationally representative survey data from 2012, 2018, and 2024.

  • Inpatient satisfaction was shaped mainly by service quality

    What the study found

    The study found that overall inpatient satisfaction was mainly linked to how patients rated five core service dimensions: medical technology, doctor-patient communication, environmental factors, medical processes, and medical costs. Among these, satisfaction was highest for medical processes and technology and lowest for costs.

    Why the authors say this matters

    The authors conclude that inpatient satisfaction is affected by multiple factors. They suggest that improving doctor-patient communication, the hospital environment and facilities, and the medical service process may help improve inpatient services.

    What the researchers tested

    The researchers conducted a cross-sectional survey of 433 inpatients at the Fifth Affiliated Hospital of Wenzhou Medical University in China. They used stratified random sampling and analyzed satisfaction with a Kruskal-Wallis test and ordered logistic regression, adjusting for sociodemographic variables.

    What worked and what didn't

    The mean overall satisfaction score was 4.49. Univariate analysis found payment method to be a significant correlate, but exploratory univariate analysis of sociodemographic factors showed few significant associations overall. Ordered logistic regression found perceived quality across the five core service dimensions to be the primary significant predictors of overall inpatient satisfaction.

    What to keep in mind

    This was a cross-sectional study from one tertiary hospital, so the summary reflects that specific setting. The abstract does not describe follow-up over time or provide additional limitations beyond the study design and location.

    • The mean overall inpatient satisfaction score was 4.49.
    • Satisfaction was highest for medical processes and technology and lowest for costs.
    • Payment method was a significant correlate in univariate analysis.
    • Perceived quality across five core service dimensions was the main predictor in ordered logistic regression.
    • The study was based on 433 inpatients at one tertiary hospital in China.

  • Dental education should build AI competencies throughout training

    What the study found

    The authors argue that dental education needs a competency framework for AI (artificial intelligence) because AI is increasingly affecting clinical workflows and processes. They propose that AI learning should be built into dental curricula from the early years through clinical training.

    Why the authors say this matters

    The study suggests that creating an AI-proficient dental workforce requires systematic educational planning. The authors conclude that AI teaching should be balanced with the core competencies required of dental professionals.

    What the researchers tested

    This article is a perspective piece rather than an experimental study. The authors discuss how AI education could be organized across preclinical and clinical years, what competencies different student levels should develop, and how assessments should be designed.

    What worked and what didn't

    The authors propose beginner-level AI knowledge and ethical considerations for early-stage students. They say senior learners should be able to use AI tools for clinical tasks and interpret AI-generated outputs, while advanced students should be able to innovate AI-driven studies and new applications for oral healthcare.

    What to keep in mind

    The abstract does not report an empirical test of the proposed framework or provide outcome data. It also does not describe specific limitations beyond noting that AI teaching must be carefully balanced with other required dental competencies.

    • The article calls for a competency framework for AI in dental education.
    • AI education should be integrated from preclinical to clinical years.
    • Early-stage students should learn basic AI knowledge and ethics.
    • Senior learners should be able to use AI tools and interpret AI outputs.
    • Advanced students should be prepared to innovate AI-driven studies and applications.
    • The authors say AI teaching must be balanced with core dental competencies.

  • Prospective teachers’ GenAI acceptance and AI literacy vary by discipline

    What the study found

    The study found that candidate teachers’ acceptance of generative artificial intelligence (GenAI) and their artificial intelligence literacy (AIL, meaning knowledge and understanding of AI) varied across several characteristics. Some differences were linked to department, grade level, AI tool use, and self-perceived proficiency.

    Why the authors say this matters

    The authors suggest that the findings help explain what influences prospective teachers’ GenAI acceptance and AIL. They also conclude that qualitative findings clarify the quantitative results.

    What the researchers tested

    The researchers used an explanatory sequential mixed methods design. They collected quantitative data from 723 prospective teachers using an information form, a GenAI Acceptance Scale, and an AIL Scale, and qualitative data from 48 prospective teachers using interviews.

    What worked and what didn't

    GenAI acceptance did not differ significantly by gender or daily internet use. It did show differences by department, grade level, AI tools used, and self-perceived proficiency. AIL differed significantly by gender, department, tool usage, and proficiency level, and scores were higher among those trained in artificial intelligence.

    What to keep in mind

    The abstract does not describe sample limits beyond the groups studied or any other methodological limitations. It also does not provide the size or direction of every difference beyond the results summarized here.

    • GenAI acceptance and AI literacy were examined among 723 prospective teachers.
    • GenAI acceptance showed no significant differences by gender or daily internet use.
    • GenAI acceptance differed by department, grade level, AI tools used, and self-perceived proficiency.
    • AIL differed significantly by gender, department, tool usage, and proficiency level.
    • Those trained in artificial intelligence had higher AIL scores.
    • Qualitative interviews were used to clarify the quantitative findings.

  • Review finds AI brings opportunities and challenges in higher education

    What the study found

    The review found that artificial intelligence (AI) in higher education brings both opportunities and challenges. It highlights benefits in teaching, learning, assessment, and administration, alongside concerns about academic integrity, ethics, psychological issues, and institutional governance.

    Why the authors say this matters

    The authors conclude that a broader, integrated approach is needed to use AI responsibly in higher education. They say this includes faculty training and use of institutional resources to benefit from AI while reducing related risks.

    What the researchers tested

    The researchers used a narrative review approach to synthesize recent research on AI in higher education. The review examined AI's effects on teaching and learning, assessments, academic integrity, ethics, psychological considerations, and institutional governance.

    What worked and what didn't

    The review says adaptive AI-based systems, intelligent tutoring platforms, and generative AI tools can improve accessibility and personalize learning, which may increase student motivation. It also says AI can improve assessment processes by providing immediate feedback and adjusting evaluations. However, heavy reliance on AI for assessment tasks raises concerns about academic integrity, cognitive offloading, and the limits of skills acquisition.

    What to keep in mind

    The abstract notes open questions about detecting AI-generated content, fake narratives in generative AI tools, bias, privacy, and environmental impact. It also says AI governance policies have not yet matured in many higher education institutions. Specific limitations of the review itself are not described in the available summary.

    • The review says AI in higher education has both positive and negative impacts.
    • It highlights benefits such as personalized learning, accessibility, motivation, and immediate feedback.
    • It raises concerns about academic integrity, cognitive offloading, and skill limits when AI is heavily used.
    • The abstract says open questions remain about AI-generated content detection, bias, privacy, and environmental impact.
    • The authors say AI governance policies are still not mature in many higher education institutions.