The survey separately targeted attending doctors and residents with similar structured questionnaires containing 27 items. These covered respondents’ demographic characteristics and AI literacy, attitudes toward the clinical use of AI, and preferences for integrating AI education into the curriculum. Responses were collected through the Sojump online platform after quick-response codes were distributed through WeChat or email.

Researchers analyzed the results in SPSS 25.0. The statistical procedures included chi-square tests for comparisons between the doctor and resident groups, reliability analysis using Cronbach’s alpha, and validity analysis with the Kaiser-Meyer-Olkin measure. The supplied study abstract does not report the individual test statistics or establish which observed group differences were statistically significant.

Attitudes toward clinical AI were generally favorable in the surveyed population. The clearest reported result concerned the possible role of AI in clinical work: 74.63% of doctors agreed that AI could partially replace doctors’ work in certain situations. This response reflects an expectation about how some tasks might be handled; the survey did not test an AI system, compare clinical performance, or measure effects on diagnostic accuracy, patient outcomes, or workflow efficiency.

Interest in further education differed between the two professional groups. A total of 78.7% of doctors said they were willing to learn more about AI, compared with 52.43% of residents. These proportions describe willingness expressed in the questionnaire rather than participation in a training program or subsequent changes in knowledge or clinical behavior.

The authors characterized the combined results as showing perceived readiness among doctors and residents to engage with AI-related material. They also interpreted the findings as evidence of demand for integrating AI education into clinical teaching. The survey was designed to examine educational preferences as well as attitudes and literacy, making curriculum integration a central reported outcome rather than an inference based only on favorable views of the technology.

Because the research was cross-sectional, it did not track participants over time and therefore cannot show that attitudes toward AI were changing. The results also come from one hospital and a defined population of attending doctors and residents, so they should not automatically be generalized to medical students, clinicians at other institutions, or healthcare systems in other regions. The study’s scope was educational needs and perceptions, not an evaluation of clinical AI performance.

Overall, the survey documents a combination of favorable attitudes, limited reported AI experience and differing levels of interest in additional learning among doctors and residents at the Tianjin hospital. Its clearest educational message is that clinicians in this sample wanted opportunities to engage with AI content, while formal coursework and direct experience remained uncommon.

Subject of Research: Cross-sectional survey of doctors’ and residents’ AI literacy, attitudes and medical curriculum preferences at Second Hospital of Tianjin Medical University.

Article Title: Doctors’ and residents’ perspectives on artificial intelligence in medicine: a survey on educational needs and curriculum integration

Article References: Hong, J., Guo, L. Y. X., Peng, Y., Chen, L., & Zhao, B. (2026). Doctors’ and residents’ perspectives on artificial intelligence in medicine: a survey on educational needs and curriculum integration. BMC Medical Education. https://doi.org/10.1186/s12909-026-10399-0

Image Credits: AI Generated

DOI: 10.1186/s12909-026-10399-0

Keywords: clinical artificial intelligence, medical education, AI literacy, curriculum integration, resident physicians, attending doctors, survey research, Tianjin Medical University, healthcare technology, Doctors, residents, perspectives

The research team, led by Yanan Xu and Li Zhang of Bengbu Medical University together with colleagues from Nanjing Drum Tower Hospital and the First Affiliated Hospital of Nanjing Medical University, conducted semi-structured interviews with eighteen nurses recruited through purposive sampling — a technique that deliberately selects participants with direct, relevant experience rather than drawing them at random. The interviews were systematically coded, with researchers organizing recurring ideas into themes, a standard qualitative method that transforms hours of conversation into a structured map of professional opinion. The study was reported according to the Consolidated Criteria for Reporting Qualitative Research, or COREQ, a 32-item checklist designed to make qualitative studies transparent and reproducible. Ethical approval came from the Ethics Committee of Gulou Hospital, Nanjing University School of Medicine, and every participant gave informed consent before taking part.

Four main themes emerged from the analysis, and together they read like a blueprint for how AI might — and might not — belong at the bedside of the dying. The first theme was recognition of AI’s potential value. The second concerned the real-world challenges that stand in the way of integration. The third captured ethical worries about privacy, fairness, and responsibility. The fourth described ethical tensions embedded in nursing practice itself. That structure matters because it separates practical barriers from moral ones, a distinction that engineers and hospital administrators often blur when they pitch new clinical tools.

On the practical side, the nurses saw AI as potentially useful in three concrete domains: workload support, symptom assessment, and communication. Anyone who has spent time in palliative care knows why these three came first. Nurses in this field juggle complex medication regimens, frequent reassessments of pain and breathlessness, and emotionally exhausting conversations with patients and families — often simultaneously. An algorithm that could automate documentation, flag subtle changes in a patient’s symptom trajectory, or help draft clear explanations for anxious relatives would free up the scarcest resource in end-of-life care: the nurse’s undivided attention. Symptom assessment is a particularly promising target, because palliative symptoms such as pain, dyspnea, and delirium fluctuate in ways that are easy to miss between rounds, and machine learning models excel at detecting patterns in longitudinal data that humans overlook.

Yet the nurses were equally clear-eyed about the obstacles. They questioned whether current AI systems could adapt to the realities of palliative care, where patients’ conditions change rapidly and where the goals of care often shift from cure to comfort. They also flagged a skills gap: AI literacy among nursing staff is uneven, and a tool that requires technical fluency the workforce does not possess risks becoming shelfware or, worse, a source of error. This concern echoes a broader finding in health informatics — that the success of clinical AI depends less on the sophistication of the model than on the fit between the model and the clinical workflow, and on the training of the people expected to use it.

The ethical concerns were the study’s most striking contribution, and they cluster around three concepts: privacy, fairness, and responsibility. Privacy is self-evident in a setting where patients share their fears, family conflicts, and end-of-life wishes with their caregivers; the nurses worried about where that information goes once it feeds a machine-learning system, who can access it, and whether consent obtained in a hospital corridor can truly cover algorithmic processing of the most personal data a person will ever disclose. Fairness concerns arise because AI models trained on data from certain populations can perform poorly for others — older patients, people with rare conditions, or those whose symptom expression differs from the training data — potentially compounding inequities at the very moment of life when no one can afford to be underserved.

Responsibility proved the thorniest issue of all. If an AI system misjudges a symptom or recommends an inappropriate intervention, who answers for it — the nurse who followed the recommendation, the clinician who deployed the system, the vendor who built it, or the hospital that purchased it? The nurses raised questions about accountability and autonomy that map directly onto an unresolved debate in AI governance: the problem of meaningful human control. In palliative care, where decisions carry irreversible weight, the nurses’ insistence that a human must remain in the loop is not technophobia; it is a demand that responsibility remain assignable to a person, not dissolved into an opaque statistical process.

Beneath these practical and governance concerns lies a deeper tension the nurses described in their own practice: the preservation of humanistic care. Palliative nursing is built on presence, touch, listening, and the willingness to sit with suffering that cannot be fixed. The participants worried that an efficiency-driven technology, however well-intentioned, could erode precisely the qualities that make end-of-life care humane — that a screen between nurse and patient might substitute transaction for relationship. This is the study’s viral insight, the one that should stop every AI developer mid-pitch: in this setting, the highest-value feature of any technology is that it gives nurses more time for the human work, not that it performs the human work itself.

The study’s findings arrive at a moment when healthcare systems worldwide are racing to integrate AI into clinical practice, often with governance frameworks that lag behind deployment. By documenting the views of nurses — the largest workforce in healthcare and the group in closest continuous contact with patients — the research provides evidence for future AI design, training, and governance in palliative care. The authors suggest their results can inform how developers build tools that respect the realities of end-of-life settings, how educators prepare nursing staff to use them competently, and how institutions write policies that assign responsibility clearly and protect patient privacy and fairness.

The study has limits worth noting. Eighteen nurses in a single country cannot speak for every palliative care setting, and qualitative descriptive studies capture perceptions rather than measured outcomes — the nurses’ views on AI’s potential are hypotheses to be tested, not evidence of what AI can actually achieve at the bedside. Still, as the first systematic exploration of this question in end-of-life nursing, the research fills a conspicuous gap. The message from the people who hold dying patients’ hands is ultimately a challenge to the AI industry: bring your algorithms into palliative care if you must, but design them to serve human connection, prove they are fair and private, and never ask them to replace the human presence that patients at the end of life need most.

Subject of Research: Palliative care nurses' views on AI and ethical dilemmas

Article Title: Exploring palliative care nurses’ views on the potential of AI and ethical dilemmas: a qualitative study

Article References: Xu, Y., Jing, J., Yan, Z., Ji, X., Zhou, Y., & Zhang, L. (2026). Exploring palliative care nurses’ views on the potential of AI and ethical dilemmas: a qualitative study. BMC Nursing. https://doi.org/10.1186/s12912-026-05318-z

Image Credits: AI Generated

DOI: 10.1186/s12912-026-05318-z

Keywords: artificial intelligence, palliative care, nursing, medical ethics, healthcare technology, qualitative research, Exploring, palliative, care, nurses, views, potential

The enthusiasm is notable given the demographic reality of spine care. Degenerative spine disease disproportionately affects older adults, and 42 percent of the survey respondents were over 65 years old. Yet age, the study suggests, is not the barrier many clinicians might assume. When the researchers tested associations statistically, younger age was significantly linked to greater openness to VR-based diagnostic education (p = 0.035) and to learning about surgery (p = 0.007), but openness to medical technology in general was largely independent of age, prior VR experience, and education level. Formal schooling made no measurable difference at all: highest educational degree showed no significant association with openness to VR or likelihood of use. In other words, the drivers of enthusiasm appear to be experiential — familiarity, exposure, and perceived usefulness — rather than demographic.

That distinction between general technology openness and VR-specific receptivity runs throughout the statistical analysis. Using Spearman’s rank correlations with Bonferroni correction, the team found very strong relationships between a patient’s openness to technology that teaches about diagnoses and their openness to VR for the same purpose (ρ = 0.79, p < 0.001), and similarly for learning about surgery (ρ = 0.82, p < 0.001). General tech openness also predicted the stated likelihood of trying VR education (ρ = 0.63–0.66, p < 0.001). Meanwhile, prior VR use was significantly associated with openness to VR specifically for learning about diagnoses (p = 0.046, ρ = 0.19). The pattern suggests that once a patient is comfortable with digital tools broadly, the jump to immersive technology is a small one — and that hands-on experience with VR itself adds a further increment of willingness.

When patients rated specific potential benefits on a five-point Likert scale, the most compelling use case was procedural rehearsal. Nearly 72 percent agreed or strongly agreed that they would feel more prepared for surgery if they could watch a VR simulation of their own procedure, and a similar proportion — 61 percent — believed VR could reduce preoperative anxiety. A slim majority, about 58 percent, agreed the technology could help them understand their medical condition better than traditional verbal or written explanations. These are clinically meaningful hopes, because preoperative anxiety is not a trivial inconvenience: prior research links it to higher complication rates, more severe postoperative pain, and longer recoveries. VR, by letting patients explore a three-dimensional model built from their own MRI or CT data, promises to compress complex anatomy into an intuitive experience that a pamphlet or a hurried office conversation cannot match.

But enthusiasm is not uniform, and this is where the study makes its most distinctive contribution. Rather than treating the cohort as a single bloc of average opinions, the researchers applied an exploratory cluster analysis — partitioning around medoids (PAM) — to sort patients with similar response patterns. Because survey data mix continuous Likert ratings, categorical demographics, and binary answers, they used Gower distance, a metric designed to handle heterogeneous data types. Silhouette analysis identified three as the optimal number of clusters, and bootstrap resampling with 500 iterations confirmed the partition was reproducible, with Jaccard similarity indices ranging from 0.786 to 0.888, values generally interpreted as good-to-high stability.

The three clusters are vividly different. The first, dubbed Tech-Savvy Professionals (n = 41), comprised mostly middle-aged patients aged 31 to 65, nearly half of whom had medical work experience, and 76 percent of whom had actually used VR before. They showed the highest openness to both technology and VR, rated the likelihood of trying VR education at 8.9 out of 10, and only 15 percent worried about understanding the technology. The second cluster, Open-Minded Novices (n = 37), is perhaps the most intriguing: despite zero prior VR experience among its members, their openness scores matched the first group, with the same mean likelihood of 8.9 — though 32 percent expressed concerns about grasping the technology. The third cluster, Skeptics (n = 36), skewed heavily over 65, had the lowest familiarity, a mean likelihood of only 6.6, and 86 percent flagged difficulty understanding the technology as a concern; just 39 percent would use VR for postoperative rehabilitation.

The authors are careful about what these clusters do and do not mean. The mean silhouette width of 0.210 — below the conventional threshold of 0.25, and echoed by a value of 0.23 from an independent hierarchical clustering method — indicates limited separation between groups. Patient attitudes toward VR, they caution, likely exist along a continuum rather than as discrete subgroups, and the clusters should be treated as hypothesis-generating profiles, not validated clinical phenotypes. Still, the framework has practical implications. Open-Minded Novices may need only low-barrier exposure and guidance to become adopters, while Skeptics may require targeted, experiential interventions. Prior research on older adults supports that approach: in one study cited by the team, attitudes toward head-mounted VR shifted from neutral to positive after direct exposure.

The barriers patients named are refreshingly concrete. The most common concern was difficulty understanding the technology (42.98 percent), followed by motion sickness (24.56 percent) and time constraints (16.67 percent). Some respondents also worried that VR might displace the human element of surgical consultation — one participant put it bluntly, noting that “the human factor is far better” — while others raised accuracy, personalization, cost, and physical discomfort with headsets. The authors emphasize that successful implementation will hinge on designing experiences that are intuitive, time-efficient, comfortable, and explicitly complementary to the surgeon-patient relationship rather than a substitute for it.

Contextually, the study fills a real gap. A 2022 systematic review of VR in neurosurgery found that only about 5 percent of studies were patient-oriented, and patients remain underrepresented across the broader field of extended reality in neurosurgery, where research has concentrated on surgeon training, neuronavigation, and pain management. Earlier intervention studies — including a randomized controlled trial showing lower preoperative anxiety and higher preparedness among surgical patients given VR — hinted at the benefits, but baseline patient attitudes before implementation had been largely unmeasured. This survey, conducted in two neurosurgical offices (one suburban, one urban) with a five-to-ten-minute anonymous questionnaire, provides that missing snapshot, though the authors acknowledge limitations: the cross-sectional, single-center design, possible sampling bias from voluntary waiting-room recruitment, exclusion of 18 percent of surveys below an 80 percent completion threshold, a non-validated survey instrument, and the absence of any actual VR exposure during data collection.

The bottom line is encouraging for anyone hoping to see immersive technology in the preoperative waiting room. Patients facing spine surgery — a population that is older, often anxious, and drowning in unfamiliar medical jargon — are broadly ready and willing to put on a headset if it helps them understand what is about to happen to their bodies. The challenge ahead is not persuading patients but engineering for them: validating these exploratory response patterns prospectively, measuring hard endpoints like comprehension scores and anxiety measures after real exposure, and tailoring the experience so that the tech-savvy are challenged, the open-minded novices are guided, and the skeptics are won over one intuitive demo at a time.

Subject of Research: Patient attitudes toward virtual reality in spine surgery education

Article Title: Exploratory clustering of spine surgery patients’ attitudes toward virtual reality

Article References: Gabriel, E. D., Wu, A. Q., Sollenberger, C. H., Rani, N. S., Malhotra, N. R., Welch, W. C., Ozturk, A. K., Shin, J. H., Godzik, J., & Judy, B. F. (2026). Exploratory clustering of spine surgery patients’ attitudes toward virtual reality. Frontiers in Virtual Reality, 7, Article 1948670. https://doi.org/10.3389/frvir.2026.1948670

Image Credits: AI Generated

DOI: 10.3389/frvir.2026.1948670

Keywords: virtual reality, spine surgery, patient education, cluster analysis, neurosurgery, preoperative anxiety, patient attitudes, survey study, healthcare technology, Gower distance, partitioning around medoids, technology acceptance

The appeal of contactless monitoring is easy to understand. Traditional hospital instrumentation, from finger-clip pulse oximeters to adhesive electrode patches and inflatable blood pressure cuffs, works well enough in controlled clinical settings, but it is uncomfortable for long-term wear, can irritate fragile skin in elderly and neonatal patients, and requires constant replacement and maintenance. Non-contact systems sidestep all of this by capturing physiological information at a distance. Cameras read the subtle color shifts in facial skin caused by pulsing blood flow. Radar transmitters bounce millimeter-wave signals off the chest and detect the millimeter-scale movements of breathing and heartbeat. Wi-Fi routers measure how the human body perturbs ambient wireless channels. Acoustic sensors and infrared detectors add further modalities, and modern machine learning pipelines fuse these raw signals into usable clinical estimates.

What distinguishes the new review is its system-level perspective. Previous surveys tended to examine a single sensing technology, such as radar or computer vision, or focused narrowly on signal processing algorithms. The authors instead characterized every included study across five dimensions: sensing modality, system architecture, the physiological parameter being targeted, the validation setting, and reported performance. They also evaluated methodological quality and, critically, asked how often privacy-preserving mechanisms had actually been built into the systems, an unusually honest question for a field whose primary instruments are cameras and radio transmitters pointed at people in intimate moments of illness.

The results reveal a field of striking imbalance. Radar and vision-based systems dominate the literature, together accounting for roughly three-quarters of all studies analyzed. These two technologies have matured fastest because their underlying physics and computer vision toolchains are comparatively well developed, and because both can plausibly reach the accuracy thresholds needed for clinical triage. By contrast, approaches based on sound, Wi-Fi sensing, embedded sensors integrated into furniture, and mixed-modality combinations remain at an earlier stage of validation, with fewer studies, smaller cohorts, and less standardized evaluation.

Even more striking is the concentration on two vital signs. Respiration rate appears in 83 percent of the reviewed studies, and heart rate in 67 percent, a significant overlap given that many systems extract both from the same chest-motion or skin-color signal. No other physiological parameter exceeds 11 percent of the corpus. Blood pressure, oxygen saturation, and body temperature, all staples of routine clinical assessment, remain largely unaddressed by contactless methods. The reason is physics as much as preference: breathing and heartbeat produce measurable mechanical displacement and periodic optical change that sensors can resolve with modest hardware, whereas blood pressure and SpO₂ estimation from a distance requires subtle modeling of vascular properties that current remote techniques capture only indirectly and with limited reliability.

Perhaps the most sobering finding concerns measurement itself. The review found that heterogeneous reporting metrics across studies prevent direct statistical comparison of performance between modalities. One radar study might report mean absolute error on healthy volunteers at one meter; a vision study might report correlation coefficients on hospitalized patients at three meters under different lighting. Without standardized datasets, common evaluation protocols, and agreed-upon error metrics, clinicians and hospital administrators have no rigorous way to compare a radar system against a camera-based alternative, and the field’s headline accuracy numbers remain effectively incommensurable. This absence of benchmarking, the authors argue, is now a primary brake on clinical translation.

When it comes to deployment, the review’s central message is that context should drive technology selection, not laboratory accuracy rankings. Radar excels in hospital wards and intensive care units, where patients may be covered by blankets, moving restlessly, or positioned unpredictably, conditions that defeat optical systems but not penetrating radio waves that measure motion through fabric and darkness. Vision and acoustic systems, meanwhile, suit structured clinical spaces with stable lighting and predictable geometry, such as examination rooms and sleep laboratories, where a camera’s rich spatiotemporal information can be fully exploited. Embedded sensor designs, built into eldercare beds and chairs, offer a middle path: technically not remote, but requiring no wearable and no user cooperation, making them practical for dementia care and nursing homes where continuous compliance cannot be assumed.

For settings where the stakes are highest, the review points to mixed radar and vision systems as the emerging precision tool. By fusing the robustness of radar with the richness of optical sensing, hybrid architectures can provide redundant, mutually validating measurements suitable for neonatal monitoring and postoperative care, where a single missed apnea event or unrecognized desaturation can be catastrophic. The trade-offs are real: greater hardware complexity, harder signal fusion, and higher cost. But as both sensing strands mature, the fusion approach increasingly looks like the endpoint toward which precision-critical contactless monitoring is converging.

The review does not shy away from the field’s persistent failures. Signal reliability in uncontrolled environments remains the fundamental technical challenge: ambient light changes, multipath radio reflections, other people moving through the scene, pets, fans, and vibrating beds all contaminate the weak physiological signals these systems hunt for. Privacy and ethical concerns compound the problem, since cameras and radar implicitly surveil patients, families, and sometimes entire rooms, raising questions about consent, data ownership, and secondary use that few current systems adequately address. Finally, poor integration with existing healthcare infrastructure, electronic health records, nursing alarm systems, and clinical workflows, means that even accurate prototypes often die at the pilot stage, unable to slot into the operational reality of hospitals.

As a forward-looking contribution, the authors propose a conceptual federated, edge-based learning framework, explicitly presented as a design proposal rather than a validated system. The core idea is compelling: instead of shipping raw camera and radar data to central servers, processing would occur at the edge device itself, with only model updates, not identifiable physiological data, shared across institutions through federated learning. Such an architecture would simultaneously address privacy, reduce bandwidth, and allow hospitals to pool learning without pooling patient data. Combined with the review’s other prescriptions, standardized open datasets, unified evaluation metrics, and co-design with clinical end users, this roadmap suggests that contactless vital-sign monitoring may finally be approaching the inflection point where it leaves the demonstration stage and becomes an invisible, ambient layer of everyday clinical care.

Subject of Research: Non-contact medical IoT systems for monitoring physiological biomarkers

Article Title: A systematic review of non-contact medical IoT systems for physiological biomarker monitoring

Article References: Banik, B., Ghose, S., Nahiun, A. H., Lin, Y., Shalaginov, A., & Ghose, D. (2026). A systematic review of non-contact medical IoT systems for physiological biomarker monitoring. Frontiers in Medical Technology, 8, Article 1917332. https://doi.org/10.3389/fmedt.2026.1917332

Image Credits: AI Generated

DOI: 10.3389/fmedt.2026.1917332

Keywords: non-contact monitoring, medical IoT, radar sensing, vision-based sensing, respiration rate, heart rate, vital signs, federated learning, edge computing, patient monitoring, healthcare technology, PRISMA review

The research team, led by Liping Wang, Manhua Nie, Yanchao Xiao, and corresponding author Sai Liu, recruited nursing students from ten universities spanning seven provinces in China between May and June 2025. Using convenience sampling, they collected data through a battery of validated instruments: a self-designed demographic questionnaire, the General Attitudes Towards Artificial Intelligence Scale, the Artificial Intelligence Literacy Scale, and the Artificial Intelligence Anxiety Scale. The final analytical sample comprised 455 nursing students, a substantial cohort drawn from a wide geographic spread that lends the findings considerable weight within the constraints of the study design. All data were analyzed using SPSS Statistics 26.0 and AMOS 29.0, the latter allowing the team to construct and test structural equation models that could probe the relationships among literacy, anxiety, and attitudes simultaneously rather than one pair at a time.

The results paint a nuanced picture of a generation caught between enthusiasm and unease. Nursing students in the sample reported moderate levels of both AI literacy and AI anxiety, while generally maintaining positive attitudes toward artificial intelligence overall. That combination—middling knowledge, middling worry, and a broadly favorable outlook—may reflect the reality of contemporary nursing education, where AI is increasingly visible in coursework and clinical discussion but has not yet become a fully integrated pillar of the curriculum. The students, in other words, are neither technophiles nor technophobes; they occupy a middle ground where their attitudes remain open but their confidence and their fears are both very much alive.

The most consequential finding, however, lies in the mediation analysis. When the researchers modeled the pathways between variables, they found that AI anxiety statistically mediated the association between AI literacy and attitudes toward AI. In practical terms, this means the relationship between knowing about AI and feeling good about AI is not a simple straight line. Higher AI literacy was significantly associated with more positive attitudes, but part of that association appeared to flow through reduced anxiety. Students who understood more about how artificial intelligence works tended to fear it less, and that lower anxiety, in turn, contributed to a more welcoming stance toward the technology. Anxiety, in this model, is not merely a background emotion—it is a mechanism through which education exerts part of its effect.

This mediation structure carries real theoretical weight. In psychological research, a mediator explains how or why one variable influences another, and identifying one changes the design of interventions. If literacy influenced attitudes only directly, the prescription would be straightforward: teach more AI content. But because anxiety sits in the pathway, the study’s authors argue that nursing education may benefit from a dual-path approach—one designed to simultaneously enhance cognitive competencies, meaning AI literacy, and to manage affective responses, meaning AI anxiety. A curriculum that builds technical understanding while explicitly addressing students’ fears about job displacement, clinical error, or professional obsolescence could, according to this model, produce more durable gains in acceptance than either strategy alone.

The technical machinery behind the conclusion deserves attention. Structural equation modeling, implemented in AMOS 29.0, allows researchers to test whether a hypothesized network of causal-style pathways is consistent with observed data, and the team supplemented their primary analysis with a stepwise model modification procedure documented in their supplementary materials, including fit indices and the rationale for each adjustment. The study also reported complete subdimension and total score results for the full sample of 455 participants, and the authors followed the STROBE checklist for reporting cross-sectional research. These methodological guardrails matter because mediation claims are easy to overstate; the authors are careful to describe the mediation as statistical, which is the appropriate caution for cross-sectional data collected at a single point in time.

That caution points to the study’s principal limitation. Because the data are cross-sectional, the direction of the pathways cannot be definitively established. It is plausible that students with more positive attitudes seek out AI knowledge, or that anxiety suppresses the motivation to learn about AI, creating a feedback loop rather than a one-way street. Convenience sampling also limits generalizability, since students who volunteer for surveys about AI may differ systematically from those who decline. The authors acknowledge these constraints implicitly through their careful language, and the field will need longitudinal and experimental designs—perhaps curricular interventions that track literacy, anxiety, and attitudes across an entire degree program—to confirm the causal architecture the mediation model implies.

Even with those caveats, the stakes for nursing are high. Artificial intelligence is widely described as a key driver of the Fourth Industrial Revolution, and its impact on healthcare is accelerating through diagnostic support tools, predictive analytics, intelligent monitoring systems, and administrative automation. Nursing students’ acceptance and readiness to use these systems are crucial for successful integration, because nurses are the professionals who will interact with AI-enabled tools most frequently at the bedside. A workforce that understands AI but fears it may hesitate at the point of care; a workforce that neither understands nor trusts it may resist adoption outright. The study’s framing of perceptions as vital preparation for the intelligent healthcare era reflects this operational reality.

The emotional dimension of technological adoption has often been treated as secondary in health professional education, where competencies and knowledge checklists dominate. This study adds to a growing body of evidence that affect is not noise in the adoption equation but a signal worth measuring and addressing. AI anxiety among students may stem from many sources—uncertainty about future employment, unfamiliarity with algorithmic decision-making, high-profile narratives about automation replacing human work—and the finding that it channels the relationship between literacy and attitudes suggests that ignoring it could undercut even well-designed technical training. Educators who pair AI content with opportunities to discuss fears openly, practice with real tools, and build confidence through supervised exposure may be acting directly on the mechanism the study identified.

The research, conducted by a team based in the Clinical Nursing Teaching and Research Section and the Department of Kidney Transplantation at The Second Xiangya Hospital of Central South University, received ethical approval from the hospital’s Ethics Committee and was funded in part by the Natural Science Foundation of Hunan Province and the Hunan Nursing Association’s Young Talent Initiative. Published open access in BMC Medical Education, it offers educators across nursing and the wider health professions a concrete, testable framework: measure literacy, measure anxiety, and design curricula that move both. As AI continues its advance into clinical medicine, the study suggests that the battle for acceptance will be fought not only in lecture halls and skills labs but in the emotional lives of the students training to become the next generation of caregivers.

Subject of Research: The mediating role of AI anxiety in the relationship between AI literacy and attitudes toward artificial intelligence among nursing students

Article Title: The mediating role of AI anxiety between AI literacy and attitudes towards AI in nursing students: a cross-sectional study

Article References: Wang, L., Nie, M., Xiao, Y., & Liu, S. (2026). The mediating role of AI anxiety between AI literacy and attitudes towards AI in nursing students: a cross-sectional study. BMC Medical Education, 26(1), Article 1657. https://doi.org/10.1186/s12909-026-10522-1

Image Credits: AI Generated

DOI: 10.1186/s12909-026-10522-1

Keywords: artificial intelligence, AI literacy, AI anxiety, nursing students, nursing education, mediation analysis, cross-sectional study, structural equation modeling, healthcare technology, medical education, China, attitudes toward AI