Background:Physician attire is associated with patient trust and confidence, yet it has rarely been examined in urology or in the surgical context in which the physician is seen. Aim:To examine how physician attire and the displayed surgical context are associated with patients' initial perceptions of urologists using standardized, artificial intelligence (AI)-generated images. Study Design:Cross-sectional study. Methods:A total of 220 adult urological patients rated 10 standardized AI-generated physician images, presented in the same fixed order, using 11 5-point Likert items and one binary acceptance item. Five images varied by attire (casual, business suit, suit with a white coat, scrubs, and scrubs with a white coat), and five varied by surgical context (neutral, open, endourological, laparoscopic, and robotic). Images were matched to each participant's stated physician-gender preference; the male-image stream (n = 189) formed the primary analysis, and the female-image stream (n = 31) formed a secondary analysis. Within-subject comparisons used the Friedman test with Bonferroni adjustment across outcomes, Page's trend test for the ordered attire hypothesis, and Wilcoxon signed-rank tests. Treatment acceptance was modeled using generalized estimating equations (GEEs) with an exchangeable working correlation. Results:Ratings varied across the fixed, ordered attire sequence for every perception item (Friedman tests, all Bonferroni-adjusted p < 0.001); Page's trend tests for the two primary outcomes were also significant (both Bonferroni-adjusted p < 0.001). Because the attire category was perfectly confounded with presentation position, this pattern cannot be attributed to attire independently of order. For male images, mean trust increased from 2.98 ± 1.34 (casual clothing) to 4.26 ± 0.81 (scrubs with a white coat), and choice likelihood increased from 3.07 ± 1.37 to 4.29 ± 0.84; observed binary treatment acceptance increased from 56.1% to 97.4%. In the adjusted GEE model, acceptance was higher at position 5 (scrubs with a white coat) than at position 1 (casual clothing) [odds ratio, 39.1; 95% confidence interval (CI), 13.1-116.7; adjusted risk difference, +40.3 percentage points; 95% CI, 32.6-48.0]. By contrast, only leadership and attractiveness remained significant among the surgical context comparisons after adjustment, and all context effect sizes were small (Kendall's W ≤ 0.03). Because the context images also differed in several unintended visual respects beyond the operative setting, this comparison is restricted to the specific images shown. Conclusion:In this standardized, image-based survey, ratings varied systematically across the fixed, ordered image sequence, with the largest and most consistent differences appearing across the images intended to vary in physician attire. Because the attire category was perfectly confounded with presentation position, this association cannot be attributed to attire alone. Associations with the displayed surgical context were small under the conditions tested and are further limited by unintended differences among those images beyond the operative setting.
Background:Lymphovascular invasion (LVI) is a critical prognostic factor in invasive breast cancer; however, reliable preoperative prediction remains challenging because of the lack of non-invasive and accurate assessment tools. Ultrasound-based radiomics and deep learning have shown promise, but conventional single-modality approaches often fail to capture intratumoral heterogeneity, thereby limiting their predictive performance and clinical interpretability. Aims:To develop and validate a non-invasive ultrasound-based model that integrates habitat radiomic features with deep learning features for the preoperative prediction of LVI in invasive breast cancer and to evaluate whether this model could enhance the diagnostic performance of radiologists. Study Design:Retrospective multicenter diagnostic study. Methods:A total of 1,096 patients from three institutions were included. Tumors on routine B-mode ultrasound were partitioned into three habitats using unsupervised k-means clustering (k = 3). Habitat2, identified as a high-risk subregion associated with LVI, was used for selective radiomic feature extraction. In parallel, deep learning features from all habitat subregions were extracted using a pretrained ResNet50 model. Radiomic and deep learning features were fused at the feature level, and a LightGBM classifier was trained and evaluated in an internal validation cohort and two independent external cohorts. Model performance was compared with that of the Habitat2 radiomics-only and ResNet50-only models using the DeLong test and with the performance of unaided senior and junior radiologists. Biological interpretability was assessed through correlation analysis with CD31-stained microvessel density. Results:The fusion model achieved areas under the receiver operating characteristic curve (AUCs) of 0.918 (training), 0.898 (internal validation), 0.890 (External Cohort 1), and 0.905 (External Cohort 2), significantly outperforming the Habitat2 radiomics-only and ResNet50-only models (DeLong test: fusion vs. Habitat2 radiomics, all p < 0.001; fusion vs. ResNet50, p < 0.001, 0.021, 0.029, and 0.026, respectively). Habitat-specific radiomic and deep learning features showed weak-to-moderate correlations with CD31-stained microvessel density, supporting the biological interpretability of the model. Compared with radiologists, the fusion model achieved higher diagnostic performance than unaided senior and junior radiologists in both external cohorts, and model assistance substantially improved the performance of junior radiologists (AUCs up to 0.865 and 0.905). Conclusion:This interpretable habitat-deep learning fusion approach based on routine ultrasound provides accurate preoperative LVI prediction with robust generalizability across multiple centers and suggests the potential to enhance clinical decision-making by improving radiologist diagnostic performance, particularly for less-experienced radiologists.
Artificial intelligence (AI) is increasingly being investigated and, in selected clinical settings, implemented to support diagnosis, triage, and workflow optimization. Although these systems have the potential to improve access, consistency, and efficiency, they may also reproduce or amplify health inequities when bias is introduced during development, evaluation, implementation, or postdeployment use. This clinician-oriented narrative review adopts a practical lifecycle approach to explain how algorithmic unfairness becomes clinically relevant, how clinicians can recognize it, and how institutions can mitigate its impact. We first outline the ethical, clinical, and mathematical dimensions of fairness. We then examine fairness risks and sources of bias across six stages of the healthcare AI lifecycle: problem formulation, data generation, model development, evaluation, implementation, and postdeployment monitoring and governance. Key mechanisms include biased proxy outcomes, unrepresentative or error-prone data and labels, model shortcut learning, hidden stratification, distribution shift, and human-AI interaction effects (e.g., automation bias and alert fatigue), all of which can create feedback loops and contribute to fairness drift over time. For each stage, we identify clinician-facing red flags and practical mitigation strategies, including defining clinically meaningful outcomes, using representative and well-documented datasets, conducting subgroup-stratified evaluations, performing external and prospective validation, justifying decision thresholds, implementing safeguards for human-AI interactions, and maintaining continuous postdeployment monitoring, including postmarket surveillance for regulated medical devices. Fairness cannot be ensured through a single metric, publication, regulatory clearance, or one-time validation. Instead, equitable healthcare AI requires transparent design, rigorous evaluation, local governance, and ongoing monitoring across diverse populations, clinical sites, devices, workflows, and time. Fairness should therefore be regarded as a continuous clinical and institutional responsibility rather than a downstream technical consideration.
Nailfold videocapillaroscopy (NVC) provides direct, non-invasive access to the peripheral microcirculation and has become a central tool in adult rheumatology. Its primary clinical value lies in the evaluation of patients presenting with signs of Raynaud phenomenon (RP) and scleroderma-spectrum disorders. In particular, NVC facilitates the differentiation between primary and secondary (scleroderma-related) RP. The characteristic scleroderma pattern observed on NVC, including giant capillaries, capillary loss, microhemorrhages, and architectural disorganization, supports the diagnosis of scleroderma-spectrum disorders. International standardization efforts have further improved the reliability and interpretability of these findings, particularly in adult populations with RP and systemic sclerosis (SSc). In children, the clinical utility of NVC is now well established for similar indications; however, interpretation requires age-specific reference data. Pediatric capillary density, diameter, and length vary throughout childhood, and minor findings such as tortuosity, crossing, and isolated microhemorrhages may also be observed in healthy individuals. Without normative pediatric data, these features may be overinterpreted as pathological findings. Beyond RP and juvenile SSc, evidence supporting the use of NVC is accumulating across a range of pediatric rheumatic diseases. NVC abnormalities have been reported in juvenile dermatomyositis (JDM), mixed connective tissue disease, childhood-onset systemic lupus erythematosus, juvenile Sjögren disease, and juvenile localized scleroderma. Among these conditions, JDM demonstrates the strongest association between NVC abnormalities—including reduced capillary density, dilatation, abnormal morphology, and microhemorrhages—and disease activity. In contrast, findings in juvenile idiopathic arthritis remain non-specific. More recent studies involving pediatric patients with Behçet disease, psoriatic arthritis, and certain autoinflammatory and vasculitic disorders suggest that NVC may detect subtle microvascular alterations; however, disease-specific patterns have not been established outside the scleroderma spectrum. Selected non-rheumatologic conditions, particularly diabetes mellitus, severe acute respiratory syndrome coronavirus 2 infection, multisystem inflammatory syndrome in children, and long coronavirus disease, further highlight the potential of NVC to reflect systemic endothelial dysfunction and non-specific microvascular injury, although these applications remain exploratory. Despite its promise, several limitations continue to restrict the broader clinical application of NVC. Pediatric reference data remain heterogeneous, scoring systems are largely adapted from adult protocols, and longitudinal evidence linking capillaroscopic changes to disease progression, treatment response, or long-term outcomes remains limited. Standardized image acquisition and reporting, multicenter cohort studies, and validation of automated image-analysis techniques are essential next steps for translating descriptive observations into clinically actionable information. This review summarizes the contribution of NVC to pediatric rheumatology and explores potential areas for future expansion. The strongest indications remain RP and scleroderma-spectrum disorders, whereas the most promising emerging application is in JDM, where microvascular changes correlate with disease activity. Across other pediatric rheumatic diseases, NVC currently serves as a complementary descriptive tool for vascular phenotyping rather than a standalone diagnostic modality. With the development of age-specific reference standards, harmonized protocols, and longitudinal validation studies, NVC has the potential to evolve from a descriptive imaging technique into an integral component of risk stratification and disease monitoring in pediatric rheumatology.
Background:Admission-based risk stratification tools are limited for hospitalized patients with fibrotic interstitial lung disease (F-ILD). Aims:To develop and externally validate admission-based machine-learning models for predicting mechanical ventilation (MV), 30-day and 3-month mortality, and long-term all-cause and cause-specific mortality in hospitalized patients with F-ILD. Study Design:Multicenter retrospective cohort study. Methods:This study included hospitalized adults with F-ILD from two tertiary hospitals in China. Clinical characteristics and laboratory test results obtained within 24 hours of admission were used as candidate predictors. Machine-learning models were developed to predict MV, 30-day and 3-month mortality, and long-term all-cause and cause-specific mortality, with internal testing and independent external validation. Model performance, clinical utility, and interpretability were evaluated. Results:A total of 1,272 patients were included (derivation cohort, n = 1,006; external validation cohort, n = 266). In the external validation cohort, the models demonstrated robust discrimination for MV [area under the curve (AUC), 0.913], 30-day mortality (AUC, 0.926), and 3-month mortality (AUC, 0.813). For long-term all-cause mortality, the final model achieved a C-index of 0.768, with time-dependent AUCs of 0.811 at 12 months and 0.768 at 24 months. The corresponding values for cause-specific mortality were 0.768, 0.815, and 0.761, respectively. Lactate dehydrogenase (LDH), neutrophil-to-lymphocyte ratio (NLR), and prognostic nutritional index (PNI) emerged as key predictors of MV and short-term mortality, whereas long-term mortality risk was additionally associated with older age, male sex, smoking history, an idiopathic pulmonary fibrosis phenotype, and lower albumin-to-globulin ratio and platelet-to-white blood cell ratio. Conclusion:Admission-based models demonstrated strong discriminatory performance for predicting MV and mortality in hospitalized patients with F-ILD. Elevated LDH and NLR levels, together with lower PNI, characterized a high-risk profile for acute deterioration and death. Local recalibration may be necessary before implementation in new clinical settings.
Background:The relationship between obstructive sleep apnea (OSA) and coronary artery disease (CAD) remains controversial. Although observational studies have suggested that OSA is associated with an increased risk of coronary events, substantial confounding and neutral findings from treatment trials have raised uncertainty regarding which aspects of OSA are most strongly linked to CAD risk. Aims:This study aimed to examine the association between nocturnal hypoxemia, assessed using the oxygen desaturation index (ODI), and incident CAD. Study Design:Prospective cohort study. Methods:Adults referred for evaluation of suspected OSA within the sleep apnea patients in Skaraborg study were prospectively followed for the occurrence of incident CAD. Associations between nocturnal hypoxemia, measured using the ODI, and incident CAD were investigated. Apneahypopnea index (AHI) categories were also analyzed for comparison. Multivariable Cox proportional hazards regression models were constructed, including models that evaluated AHI and ODI separately as well as simultaneously to determine their independent associations with CAD risk. Results:The analytic cohort included 2,902 adults (1,012 women and 1,890 men) with a median follow-up duration of 8.7 years, during which 111 incident CAD events were identified. In fully adjusted Cox regression analyses, participants in the highest ODI category (≥ 30 events/h) had a significantly greater risk of incident CAD than those in the lowest category (hazard ratio, 1.99; 95% confidence interval, 1.08-3.65). In contrast, AHI categories were not independently associated with incident CAD. Moreover, when AHI and ODI were simultaneously included in the same model, ODI remained independently associated with incident CAD. In analyses restricted to adults with OSA, the association between ODI and CAD was attenuated and did not reach statistical significance, although the direction and magnitude of the association remained generally consistent with those observed in the primary analysis. Conclusion:Among adults referred for OSA evaluation, nocturnal hypoxemia assessed using ODI, rather than event-based OSA severity, was associated with incident CAD. However, the absence of independent associations when both ODI and AHI were considered simultaneously suggests substantial overlap between these OSA severity metrics. These findings suggest that hypoxemia-related parameters may provide additional value for coronary risk stratification.
Background: Differential expression (DE) analysis of RNA sequencing (RNA-Seq) data are cornerstone of transcriptomic research. Widely used statistical frameworks are primarily optimized to detect monotonic mean shifts between conditions and may therefore overlook genes or microRNAs whose disease association arises at both low and high expression levels. Such non-monotonic patterns, referred to here as improper expression profiles, may reflect biologically relevant heterogeneity but remain difficult to identify using standard tools. Aims: To evaluated whether receiver operating characteristic (ROC)-based indices, specifically the generalized area under the curve (gAUC) and the length of the ROC curve (LROC), can support exploratory screening and prioritization of improper expression profiles in RNA-Seq data, as a complement to conventional DE methods. Study Design: Methodological study. Methods: Using simulated negative binomial count data, we compared DESeq2, classical AUC (cAUC), gAUC, and LROC across varying sample sizes and dispersion levels, focusing on improper expression profiles. Performance was summarized using true positive rate and positive predictive value under ranking-based feature selection, including a one-shot benchmark operating point (available only in simulations) and sensitivity analyses across selection sizes. The methods were also applied to a publicly available CC miRNA dataset using heuristic post-hoc screening rules informed by simulation diagnostics. Results: cAUC was largely insensitive to improper expression patterns. DESeq2 performed robustly for conventionally differentially expressed features but recovered a smaller fraction of simulated improper profiles under ranking-based selection. Across simulation scenarios, gAUC showed the highest and most stable recovery of improper profiles, whereas LROC provided complementary signal under low-to-moderate dispersion but degraded under extreme overdispersion. In the CC dataset, ROC-derived indices identified candidate improper miRNAs that were not prioritized by DESeq2, and several top candidates had literature support consistent with biological plausibility. Conclusion: gAUC, supported by LROC as an auxiliary index, provides a practical ROC-based screening extension to standard RNA-Seq workflows. Because these indices are applied using heuristic thresholds without controlled error rates, the resulting candidates should be interpreted as exploratory prioritization and require independent validation.
Kounis syndrome is defined as the occurrence of acute coronary events in the setting of allergic, hypersensitivity, or anaphylactic reactions. It is mediated by mast cell activation and the interaction of inflammatory cells, including T lymphocytes and macrophages. This process leads to the release of multiple inflammatory mediators, such as platelet-activating factor, histamine, neutral proteases (tryptase and chymase), arachidonic acid metabolites, cytokines, and chemokines. Kounis syndrome represents a unique form of acute vascular disorder that may involve not only the coronary arteries but also peripheral, cerebral, and mesenteric vessels as well as the venous system. Contrast media are widely used in diagnostic imaging to enhance visualization and characterization of pathological conditions. These agents can be administered via several routes, including oral, intravenous, intra-arterial, or rectal administration. Although most hypersensitivity reactions to contrast media are mild to moderate, severe complications such as anaphylaxis, cardiac arrest, and Kounis syndrome may occur. In particular, contrast media-induced Kounis syndrome has been associated with significant clinical consequences, including an increased risk of life-threatening cardiac events. This narrative review aims to summarize current evidence regarding contrast media-related adverse effects, with a focus on hypersensitivity reactions, Kounis syndrome, and associated cardiovascular complications. Emphasis is also placed on preventive strategies, including the importance of obtaining a detailed patient history of prior hypersensitivity reactions prior to contrast administration, to reduce the risk of recurrence and severe outcomes.
