Search Results (161)

While variable-latency AI dialogue processing proceeds asynchronously, the VR state continues to be updated frame by frame. In this study, we implemented “guard-mediated LLM-VR scene control” for breastfeeding support training, in which LLM-derived ActionClass and PhaseCandidate are not used to update the phase directly, but are instead provided as inputs to phase-specific deterministic guards together with the VR state and the ActionClass-mediated mother-avatar state on the Unity side. We conducted human-in-the-loop sessions with 20 nursing students, post-study guard/Turn-ID tests, a 2-expert semantic evaluation of 639 cases, and state-binding and local guard sensitivity analyses. There were no failures preventing session continuation, and the median of the participant-specific median latencies was 2.61 s. Phase guards matched a separately implemented specification-based oracle in 1600 out of 1600 cases, and the Turn-ID gate eliminated cross-turn stale results. LLM-expert strict agreement was 74.5%/77.8% for ActionClass and 16.9%/35.2% for PhaseCandidate. The local outcome divergence in cases where semantic labels differed between the two experts was 85.2% in Phase 1, 1.9% in Phase 2, and 56.8% in Phase 3; the impact of semantic variability differed across phases. In particular, the current topology, which uses PhaseCandidate as a standalone transition-enabling disjunct, requires reevaluation. This study is a technical pilot and mechanism-specific characterization; it does not evaluate educational effectiveness, clinical efficacy or safety, or superiority over alternative architectures. Full article

Spiritual care is summoned by suffering—by ‘wounds’ that are psychological, relational, spiritual, embodied, and political at once. As disciplines continue to be entrenched and siloed amidst the growing sociopolitical tensions and global crises in a decolonial age, what is at stake is the ability to care for each suffering Other in their complexity. This article offers a selective genealogy and relational integration of psychoanalysis and spiritual care (in a decolonial age). Against both clinical reductionisms and theological certainty, we argue for renewed dialogue between psychoanalysis and spiritual care through a relational integration framework, both as an ethical stance and hermeneutic framework. We first trace how pastoral and practical theology have approached each ‘wound’ in spiritual care through clinical pastoral education, psychoanalytic pastoral care, practical–theological hermeneutics, and liberationist, intercultural, postcolonial, and decolonial turns. We then consider how psychoanalysis itself has undergone significant transformations through object relations, relational–intersubjective theories, existential–phenomenological approaches, and feminist, intercultural, liberationist, and decolonial critiques. Finally, we propose embodied, hermeneutical, developmental, and intercultural commitments for relational integration, further clarified through spiritual dwelling, seeking, and crucibles of transformation. These commitments orient practitioners toward more accountable forms of interdisciplinary formation, witness, healing, justice, and liberation with the suffering Other. Full article

African immigrant men in the United Kingdom (UK) experience forms of suffering that transcend both the clinical framework of acculturation stress and the political narrative of racial victimhood. Existing scholarship frequently overlooks how race, religion, and gender historically entangle to construct diasporic identities. This paper investigates how African immigrant men in the UK navigate the constitutive forces of race, gender, and religion in constructing and resisting diasporic identities. The paper develops “Ntu Fracture” as a theological–anthropological framework to diagnose the wounding of diasporic masculine personhood at the level of vital relationality. The framework draws upon Bantu philosophy (Ntu, ubuntu) and Yoruba cosmology (àṣẹ), placing them in critical dialogue with Fanon’s sociogeny and Butler’s concept of precarity. Utilising a qualitative desktop methodology, the findings suggest a fractured subjectivity among African men in the UK driven by three main mechanisms: (1) institutional racism as recognition-severing, (2) immigration enforcement as obligation-interruption and (3) parenting surveillance as domestic colonisation. The paper concludes that the lived experiences of African religious communities within the UK’s hostile environment stem from historical constructions of race, religion, and gender, categories experienced as unified entities that collapse conventional distinctions between the sacred and the secular. Full article

Chronic rhinosinusitis with nasal polyps (CRSwNP) is a heterogeneous inflammatory disease in which type 2 inflammation, epithelial dysfunction, and tissue remodeling determine severity, recurrence, and treatment response. Although molecular biomarkers have clarified disease endotypes, their routine use remains limited by cost, availability, and invasiveness. Nasal cytology offers a simple, repeatable, and minimally invasive method to assess—at the mucosal surface—both epithelial morphology and the dominant inflammatory infiltrate, whether neutrophilic, eosinophilic, mast cell, or mixed. Clinical-Cytological Grading (CCG) integrates the dominant cytological pattern with selected comorbidities, including asthma, allergy, and NSAID-exacerbated respiratory disease (N-ERD), into a weighted clinical-cytological framework. In the founding cohort, the highest relapse association was observed when mixed eosinophil–mast cell inflammation coexisted with asthma and N-ERD. This review discusses the rationale, clinical relevance, and translational applications of CCG in CRSwNP, addressing eosinophilic and mixed mast cell–eosinophilic inflammation, epithelial morphology, disease recurrence, difficult-to-treat phenotypes, biologic monitoring, and the operative dialogue between nasal cytology and histopathology. By linking cytological findings with selected clinical comorbidities, CCG may support biologically informed patient characterization and may prompt targeted mast cell assessment in tissue. However, its prognostic accuracy, incremental clinical value, and role in therapeutic decision-making require independent external validation. Full article

Biomedical vision–language models increasingly support image-grounded clinical dialogue, yet most deployable systems still depend on autoregressive language generation. Such systems tend to truncate answers, react poorly to length instructions, and offer no principled way to signal uncertainty when image evidence is weak. We present MedDiffVL, a biomedical vision-language model that pairs a masked language diffusion backbone with a SigLIP-2 visual encoder and a multimodal alignment pipeline that injects modality and question-type cues. Three inference-time mechanisms target the failure modes of diffusion-based generators in the clinical setting. An adaptive confidence-guided remasking rule uses a time-aware threshold and a short-window stability check to remove repetitive low-quality candidates. A clinically aware length controller selects a target length from question type, modality, and an internal uncertainty estimate. A reliability gate combines visual-evidence and answer-confidence scores to emit, hedge, or escalate a response. On VQA-RAD, SLAKE, and PathVQA, the model reaches 85.42, 92.78, and 94.91% closed-form accuracy and an overall conversation score of 53.42 against a fixed reference. Token repetition falls from 0.18 to 0.06. An ECE falls from 0.137 to 0.034, but this reflects an ECE-surrogate training loss and is not independently validated. These gains are not uniform. The closed-form gains over the prior diffusion model lie within run-to-run variance, and latency stays higher than autoregressive baselines. The main contribution is controllability and reliability-aware decoding, not higher closed-form accuracy. The results indicate that confidence-guided masked diffusion with reliability-aware decoding is a useful direction for controllable and reliability-aware clinical assistants. Full article

Background/Objectives: For over two millennia, Ganoderma lucidum has served as a traditional remedy, yet its translation into evidence-based therapy remains stymied by a persistent obstacle. Potent in vitro activities consistently fail to translate in vivo, a shortfall rooted in the poor systemic bioavailability of its signature triterpenoids and polysaccharides. We contend that the defining hurdle is no longer compound discovery but whether emerging formulation technologies genuinely overcome these barriers or merely sidestep them in ways that obscure the underlying physiology. Methods: We construct a structure–activity–formulation (SAF) prioritization framework. We critically survey a broad spectrum of delivery platforms—including lipid-based, polymeric, micellar, tellurium nanorod, and G. lucidum-derived vesicular systems—against their demonstrated capacity to improve oral exposure. Clinical assessments and regulatory considerations are integrated to anchor the analysis in translational reality. Results: Bioavailability has been systematically treated as a post hoc variable, even though the very structural features conferring bioactivity impose the steepest systemic barriers. To date, no G. lucidum nanoformulation has advanced to human trials, and the regulatory terrain for these complex natural-product nanomedicines remains uncharted. We distill three experimentally tractable propositions—testing lipid-mediated absorption, triple-helix conformational dependency, and co-delivery synergy—that offer discrete entry points for rigorous investigation. Conclusions: We argue that the field must pivot decisively from descriptive cataloging to hypothesis-driven, comparative investigation, anchored by reference-material standardization and proactive regulatory dialogue. Confronting the bioavailability bottleneck as a primary design parameter—rather than an ancillary nuisance—represents the most credible route from historical remedy to evidence-based therapeutic. Full article

Background/Objectives: Medical devices are essential to modern healthcare, but their adoption is often driven by regulatory conformity, clinical enthusiasm, procurement pressures and vendor-led innovation rather than systematic evaluation of comparative value. CE marking and related regulatory mechanisms are necessary for market access; however, they do not determine whether a device improves patient-related outcomes compared with existing alternatives, whether its benefits justify its total costs, or whether it can be implemented safely in routine care. This narrative review examines why medical devices require a dedicated life-cycle health technology assessment (HTA) approach and proposes an operational framework linking assessment to adoption, evidence generation, reassessment and disinvestment. Methods: A structured targeted search covered peer-reviewed literature and policy or institutional documents addressing HTA, medical devices, regulation, economic evaluation, real-world evidence, hospital-based HTA, procurement digital and AI-enabled devices, patient involvement and post-market reassessment. Results: Medical devices differ from pharmaceuticals through user dependence, learning curves, procedure dependence, short product life cycles, incremental modification, heterogeneous comparators, limited randomized evidence and hidden life-cycle costs. These features create clinical, economic, organizational and implementation uncertainty after market entry. The proposed model specifies six linked phases: horizon scanning and early dialogue, pre-adoption appraisal, an explicit adoption decision, controlled implementation, real-world monitoring and scheduled or trigger-based reassessment leading to continuation, scale-up, restriction, or disinvestment. Practical constraints include fragmented data infrastructure, the cost of maintaining registries and residual confounding in real-world evidence. Conclusions: Medical device HTA should move beyond one-time pre-adoption assessment toward a decision-linked life-cycle model that integrates comparative value, patient and public involvement, procurement, implementation governance, real-world evidence, version monitoring, reassessment and disinvestment. This approach can support responsible innovation, patient safety, transparent procurement and sustainable health-system value. Full article

Background: Speech models (wav2vec 2.0, HuBERT, Whisper), large language models (GPT, LLaMA), and conversational AI have expanded computational speech analysis from handcrafted acoustic features to dialogue-based neurological assessment. How well these approaches address clinical practice has not been evaluated. Methods: We conducted a narrative review searching PubMed, Google Scholar, and IEEE Xplore, supplemented by Interspeech and ICASSP proceedings. Findings are organized along three layers: acoustic-motor (voice quality, prosody, articulation), language-transcript (lexical, syntactic, semantic, and discourse analysis), and integrated multimodal-conversational (interactive dialogue systems). Traditional acoustic biomarkers provide background; the primary focus is on foundation models, LLMs, and conversational AI. Findings: Speech foundation models outperform handcrafted features on several classification tasks but degrade on severely impaired speech due to domain mismatch with healthy training data. LLMs classify transcripts and score cognitive tests, but operate on text alone and cannot access acoustic-motor information. Conversational AI can administer cognitive screening through naturalistic dialogue, but validation is limited to small single-centre feasibility studies. Prospective clinical validation remains limited. Cross-linguistic generalizability is untested for most methods. Interpretation: The field is moving toward integrated speech-language assessment, but the gap between technical capability and clinical utility remains wide. Closing it requires diverse multilingual datasets, standardized benchmarks, prospective validation, and ethical governance. Full article

Background/Objectives: Clinical documentation places a significant time burden on healthcare professionals, including in the context of home care. Large language models (LLMs) offer potential for automated note generation, but current approaches rely on static prompt templates that fail to generalize across care settings, languages, and documentation formats. This study proposes and evaluates an adaptive retrieval-augmented generation (RAG) framework that uses retrieval as a format adaptation mechanism, enabling the generation of structured clinical notes from patient–provider transcripts across various documentation formats without model fine-tuning. Methods: The proposed framework retrieves dialogue–note pairs that demonstrate the structure of specific sections, allowing the transfer of formatting knowledge during inference. Experiments were conducted on three datasets covering two languages and various documentation formats: the Japanese Visiting Nurse corpus (JP-VN), MTS-Dialog, and ACI-BENCH. Six controlled conditions were evaluated: zero-shot prompting (C1), static few-shot prompting (C2), dense retrieval (C3), random retrieval (C4), sparse BM25 retrieval (C5), and hybrid retrieval using reciprocal rank fusion (RRF) (C6). Performance metrics include structural adherence to required section headings, content quality (ROUGE-1, BLEU, BERTScore), and the number of hallucinated clinical entities per generated record. Results: Structure compliance increased from 0–37% under static conditions (C1/C2) to 91–100% under all adaptive RAG conditions (C3–C6) across all datasets. On MTS-Dialog, dense retrieval achieved the highest content quality (ROUGE-1: 0.519 vs. 0.446–0.492 for C4–C6; $p<0.001$). Hallucinated entities in JP-VN decreased from 2.73–3.58 per note (C1/C2) to 1.15–1.30 (C3–C6), an approximately 55–56% reduction. Conclusions: Adaptive RAG can improve structure compliance and reduce hallucinations in multilingual clinical note generation without dataset-specific prompt engineering or model fine-tuning. These findings support retrieval-based format adaptation as a generalizable mechanism for diverse clinical documentation contexts. Full article

Background/Objectives: Social media is an increasingly important source of pediatric health information for parents. In particular, the growing presence of pediatricians on social networks opens up new scenarios for adapting to the communication models of the younger generations, supporting information, prevention and parental support, but also with critical issues related to integration with the traditional clinical relationship. This study aimed to evaluate the use of social media for pediatric health information, the perceived usefulness of content shared by pediatricians, its impact on parental behavior and the degree of integration of this information into dialogue with the treating pediatrician. Methods: A cross-sectional observational study was conducted using an online questionnaire aimed at 453 parents of minors. The questionnaire investigated socio-demographic characteristics, frequency of following pediatricians on social media, perceived usefulness of online health information, impact on parenting behaviors, discussion of content with the treating physician, and perception of risks and the need for regulation. Associations between categorical variables were evaluated using Fisher’s exact test and multivariable logistic regression. Results: Most participants follow pediatricians on social media (74.2%) and rate online health information as very useful (81.7%). Almost two-thirds reported changing at least one health-related behavior following exposure to pediatric content on social media (65%). Greater frequency of following pediatricians was significantly associated with both higher perceived usefulness and behavioral change (p < 0.0001). Perceived usefulness emerged as the strongest independent predictor of behavioral change (OR 9.967; p < 0.001). Conclusions: Pediatric communication on social media appears to be widely used, perceived as useful and associated with parental behavior. However, such content is poorly integrated into clinical dialogue, mainly taking the form of a one-way flow of information. The results highlight the need to promote more participatory communication models and to recognize pediatricians active on social media as a distinct professional category of digital health communicators. Full article

Clinical documentation in Electronic Health Records (EHRs) remains a substantial source of administrative burden for clinicians. In this study, we evaluate a modular AI-assisted clinical documentation pipeline using two complementary approaches: (1) a controlled benchmark based on multilingual synthetic clinical dialogues, and (2) an observational analysis of real-world usage traces from routine deployments. The benchmark enables systematic comparison of ASR–LLM configurations under fully controlled conditions, using metrics for transcription accuracy (Word Error Rate and Medical WER), report-generation quality, and modeled processing cost. Within this benchmark setting, Voxtral showed the strongest ASR performance among the evaluated models, while GPT-4o and Gemini 1.5 Pro showed the strongest report-generation performance under the automated evaluation used in this study. The real-world trace analysis should be interpreted as descriptive evidence of operational use, not as prospective clinical validation or as a direct evaluation of any single benchmarked configuration. Taken together, the results support the use of this pipeline as a human-supervised draft-generation tool that still requires clinician review, local workflow evaluation, and prospective clinical validation before broader deployment. Full article

Introduction: As diabetes care becomes increasingly digitalized, stratified, and differentiated, psychosocial vulnerability risks becoming less visible within routine care and documentation. To ensure that differentiated care pathways meaningfully incorporate psychosocial stratification, vulnerability must be identifiable, documented, and revisited as part of routine clinical practice. Aim: The aim of this study is to explore how healthcare professionals identify psychosocial vulnerability in routine diabetes care and how such vulnerability is documented and followed up in the electronic medical record (EMR). Methods: This quality improvement audit with a descriptive analysis component was conducted at Steno Diabetes Center Copenhagen as part of the development of a new differentiated outpatient pathway. Healthcare professionals across disciplines submitted cases of persons with diabetes whom they considered psychosocially vulnerable. Documentation from the preceding six months was reviewed descriptively in order to find patterns of identification, documentation, care planning, and follow-up. Results: A total of 334 referrals representing 275 unique persons with diabetes were submitted. Psychosocial vulnerability extended beyond predefined high-risk categories, as 37% of identified cases did not align with any of the six vulnerability groups described in the Danish VIVE framework. Vulnerability often reflected cumulative everyday-life strain rather than formal diagnoses. While healthcare professionals demonstrated substantial relational attentiveness to psychosocial concerns, this knowledge was not consistently evident in formal documentation. Explicit care plans and longitudinal follow-up were uncommon, and psychosocial concerns were frequently documented as isolated observations rather than as part of structured ongoing care. Conclusions: Psychosocial vulnerability was frequently identified through clinical dialogue and professional judgement and often extended beyond predefined vulnerability categories. The findings highlight the importance of developing shared approaches and a shared understanding of psychosocial vulnerability across professional groups. If psychosocial stratification is to become an operational component of differentiated diabetes care, information about what burdens matter to the person must be identifiable, documented, and carried forward across encounters alongside biomedical information. Full article

Background/objective: Given the increasing complexity of the luminal breast cancer landscape, a proper characterization is required in everyday clinical practice, and the recurrence after the standard first-line treatment with CDK4/6 inhibitors with/without endocrine therapy should be managed. Method: The Hermione Exchange Educational Program was held in Milan, Italy, between September 2024 and January 2025. Two questionnaires were proposed regarding the use of targeted treatment or chemotherapy after progression from CDK4/6 inhibitors. The lecture and use cases enhanced the discussion during the workshops. Results: From the surveys, it emerged that most participants (69%) considered liver metastases at CDK4/6-inhibitor progression as a key reason to initiate chemotherapy, while lung progression influenced this choice for 50% of participants. Liver involvement guided the use of targeted therapy for 56%, and attitudes were divided on whether the duration of first-line CDK4/6 therapy should affect decisions (44% in agreement vs. 38% in disagreement). The willingness of patients to receive chemotherapy (88%) and comorbidities (81%) were significant drivers. Almost all participants agreed that both the duration of response and the molecular status were key aspects to consider when choosing a second line of therapy, along with the general clinical condition of the patient. In the lecture, tissue and liquid biopsy are considered powerful tools to describe tumor molecular features over time; such complexity should be harnessed by a close dialogue between oncologists, molecular biologists, and pathologists to optimize the therapeutic choice according to the mutational status of patients. The use cases illustrate three patients with visceral progression, non-visceral progression within 12 months, and non-visceral progression after 12 months following CDK4/6 inhibitors. Conclusion: Genomic testing should be considered at diagnosis and repeated during treatment to monitor the disease. The clinical experience acquired over the years must be integrated with new molecular knowledge. Full article

The prevention and treatment of cardiovascular disease (CVD) are undergoing a paradigm shift from a lipid-centric approach to a holistic metabolic perspective. Central to this evolution is the gut–fat–heart axis, a sophisticated three-dimensional communication network that integrates neural, endocrine, and immunometabolic signaling to regulate systemic lipid homeostasis. This manuscript systematically explores how the gut microbiota acts as a “metabolic organ” to remotely control host health through the production of bioactive metabolites and the modulation of molecular communication networks. At the physiological level, microbial products such as short-chain fatty acids (SCFAs) and modified bile acids regulate energy balance and lipid synthesis via the FXR-FGF15/19 axis and G protein-coupled receptors. Furthermore, gut hormones like GLP-1 and neuro-reflex pathways involving the vagus nerve provide rapid control over postprandial lipid clearance and feeding behavior. Conversely, pathological dysbiosis triggers the accumulation of harmful metabolites, such as trimethylamine N-oxide (TMAO) and lipopolysaccharides (LPS), which drive lipotoxicity, vascular inflammation, and “dysfunctional HDL” formation. These processes accelerate the progression of atherosclerosis, heart failure, and metabolic syndrome. Finally, the article outlines promising clinical translation strategies, including the development of TMA lyase inhibitors, next-generation probiotics, and the use of phytochemicals to reshape the microbial landscape. By decoding the molecular dialogues within the gut–fat–heart axis, this research provides a novel strategic vantage point for the integrated management of cardiovascular–kidney–metabolic (CKM) syndrome. Full article

Despite recent advances, medical question answering systems still struggle with domain-specific reasoning and data efficiency. This paper presents Med-LLaMA3, a family of medical large language models developed by parameter-efficient fine-tuning of the LLaMA-3.1 (8 billion) and LLaMA-3.2 (1 and 3 billion) architectures using quantized low-rank adaptation (QLoRA) and low-rank adaptation (LoRA) with 4-bit quantization. Beyond model training, this work contributes the following: (1) a formalized dataset curation taxonomy (source type × clinical granularity × task format) with a source-category ablation confirming that the multi-source combination drives benchmark gains beyond any single category; (2) a systematic characterization of low-rank-adaptation rank-scaling behavior for the LLaMA-3 family in the medical domain (monotonic improvement up to rank 128, with no observed plateau); and (3) statistically validated comparisons using McNemar’s test and 95% bootstrap confidence intervals. We curated a medical instruction dataset of over 1.5 million samples spanning medical examinations, clinical dialogues, and biomedical literature. Our approach trains only ∼4% of the base model’s parameters and, consistent with prior studies of parameter-efficient methods in the medical domain, achieves performance comparable to full fine-tuning at a fraction of the memory footprint. Evaluated with five in-context examples per prompt, the 8-billion-parameter model attains a mean accuracy of 75.71% across the eight medical-domain subsets of the Massive Multitask Language Understanding benchmark; improvements over the unmodified LLaMA-3.1-8B-Instruct baseline are statistically significant on the medical multiple-choice benchmark MedMCQA and, after Bonferroni correction across the eight subsets, on three subsets (Clinical Knowledge, Medical Genetics, and Nutrition), with two further subsets being significant only before correction. A structured named-entity-recognition evaluation on 100 hospital discharge summaries (macro-averaged F1 0.94; dual-annotator agreement $\kappa =0.87$) provides complementary evidence of clinical-text utility. A safety mitigation pilot shows that context-disambiguation preprocessing reduces the highest-severity abbreviation-ambiguity error rate from 30% to 10% on a 30-case held-out set. These results show that parameter-efficient fine-tuning can deliver high-performance medical large language models while training only ∼4% of the model’s parameters and reducing memory use by roughly 75%, enabling development on low-cost consumer-grade hardware. Full article