Search Results (2,795)

Background: The early diagnosis of eye fundus pathologies is crucial, as they may go unnoticed until reaching advanced stages. To offer an improved screening methodology for this purpose, the effectiveness of a conditional variational autoencoder (CVAE) based on multispectral (MS) imaging and operating from the visible to the near-infrared range (416–1213 nm) has been assessed. Methods: A total of 2040 images from 102 patients (66 females, 36 males; aged 19–91 years) were acquired with an MS fundus camera to feed a fine-tuned CVAE for classifying eye fundus as healthy or diseased. The performance of the neural network was assessed for different image resolutions and spectral ranges. Results: The proposed deep generative model showed excellent results, reaching 100% of accuracy, sensitivity and specificity for the set of MS images from 416 nm to 955 nm at maximum resolution (1757 × 1757 pixels). Other instances with different image resolutions and spectral ranges led to good classifications (accuracy between 96% and 98%, sensitivity between 92% and 98%, and specificity between 97% and 100%). The CVAE exhibited robust performance with convergence of the accuracy and loss through the different epochs for training and validation in all instances. Conclusions: This study proves that a CVAE approach based on MS imaging is a highly effective tool for diagnosing eye fundus conditions and could potentially serve as a valuable clinical support tool for screening. The approach performs remarkably well when high spatial resolution MS images ranging from 416 nm to 955 nm are used. This underscores the importance of combining spatial and spectral information, particularly of wavelengths beyond the visible range. Full article

Background: Acute confusional state, or delirium, is a common neuropsychiatric disorder in older adults, particularly in emergency departments. It is associated with high morbidity and mortality, as well as difficult detection. Its early identification is crucial to prevent complications and improve prognosis. This study aims to identify and analyze available strategies for the early diagnosis of delirium in geriatric patients in emergency departments. Methods: A literature review and meta-analysis were conducted between 2020 and 2026 using the PubMed, Scopus, Web of Science, CINAHL, and Cochrane databases (PRISMA 2020). To estimate the prevalence of patients with delirium in emergency departments, a meta-analysis was performed using Stats Direct statistical software (Version 4). Methodological quality was evaluated according to the Oxford Centre for Evidence-Based Medicine levels of evidence. Results: Eleven studies were included, demonstrating that delirium in emergency departments is highly prevalent among older adults, particularly those with dementia, polypharmacy, or functional impairment. The 4AT scale was the most frequently used screening tool, notable for its strong sensitivity. Hypoactive delirium emerged as the most common subtype and carried the poorest prognosis. Furthermore, its presence was associated with longer hospital stays, increased complications, and higher mortality rates, while a lack of training among healthcare staff continues to limit early detection. The meta-analysis detected no publication bias and revealed a 20.6% prevalence of delirium among patients in emergency departments. Conclusions: One in five patients attending the emergency department presents with delirium. Delirium in geriatric emergency care demands a structured clinical response grounded in prevention, early detection, and professional training. Integrating validated diagnostic tools and reinforcing the role of nursing professionals are key to improving patient outcomes and safety. Full article

Therapies based on mesenchymal stromal cells (MSCs) have high potential in the field of regenerative medicine due mainly to their immunomodulatory properties. However, their clinical translation is hampered by a lack of sufficiently standardised potency tests. Since macrophages constitute key mediators of the effects of MSCs, macrophage-based assays potentially provide a relevant in vitro tool for the evaluation of the activity of MSC products. This study involved the coculturing of canine adipose-derived mesenchymal stromal cells (ASCs) with macrophages derived from human THP-1 and U937 monocyte cell lines, murine RAW264.7 macrophages and primary human macrophages. The M2 polarisation was assessed following stimulation with IL-4/IL-13 in THP-1 and U937 macrophages. The mRNA expression of the pro- and anti-inflammatory markers was analysed using qPCR. The ASC transwell coculture altered the LPS-induced inflammatory mRNA expression in a strongly model- and marker-dependent manner. The U937-derived macrophages exhibited the most consistent suppression of the tested inflammatory transcripts and the RAW264.7 cells provided a practical readout for selected inflammatory markers, whereas the THP-1 macrophages evinced the suppression of TNFA but not IL1B or PTGS2 under the selected stimulation conditions. IL-4/IL-13 induced moderate but statistically non-significant changes in IL10 and TGFB1 in the U937-derived macrophages but no reproducible response in the THP-1-derived macrophages. In a subsequent U937 coculture experiment, ASC-derived paracrine factors altered selected M2-associated transcripts at specific time points. The results thus provided support for macrophage-based transcriptional readouts as an early-stage tool for comparing responder macrophage models and detecting the selected anti-inflammatory paracrine effects of canine ASCs; the U937 cells were found to be particularly suitable for the study of inflammatory polarisation and the RAW264.7 cells for the purpose of standardised screening. Full article

Background: Q&A corpora generated through university student–AI mental health support tools may reveal how physical activity (PA) and psychosocial themes are represented in support-oriented text. However, the absence of individual-level demographic metadata and the pooling of prompt and response fields limit attribution of any expression to a particular speaker, and the corpus describes a specific student population rather than a general or clinical one. Objective: This exploratory study described PA-, sport-, physical education (PE)-, body-, lifestyle-, and emotion-related patterns in a large corpus of university student–AI mental health exchanges collected through an institutional counselling platform. Methods: This study analysed 209,715 paired prompt–response records as combined exchange-level units using a BERTopic-based computational text-mining workflow. The full corpus was used for the main 18-topic model and overlapping dictionary analyses. After secondary data-quality filtering, 178,062 eligible exchanges formed the sampling frame from which a systematic sample of 10,000 exchanges was drawn for a separate complementary BERTopic and scenario-mapping analysis. The workflow used Qdrant/bge-small-zh-v1.5 embeddings, NFKC normalisation, an archived stop-word list, UMAP (n_neighbors = 15, n_components = 5, min_dist = 0.0, cosine metric, seed = 42), HDBSCAN (min_cluster_size = 300, min_samples = 10, Euclidean metric, EOM), c-TF-IDF topic representations, overlapping dictionary screens, and stability testing across seeds 42, 52, and 62. Results: A student/school/family-context lexical screen matched 83,215 exchanges (39.68%), and a broad PA/body/lifestyle screen matched 82,464 exchanges (39.32%). These overlapping indicators describe topical co-occurrence and do not establish PA behaviour or which party to the exchange produced a given term. Eighteen corpus-level themes were retained. In the 10,000-exchange analysis, 13.11% of exchanges matched a narrow movement-related expression screen, with the highest within-topic rate in the sample topic labelled emotional outburst and relaxation regulation (51.09%). Conclusions: The findings describe exchange-level lexical and topic patterns in student–AI interactions rather than actual PA behaviour, intervention delivery, clinical efficacy, or population prevalence, and they do not identify which party introduced the language. The mapping to autonomy, competence, relatedness, and emotional regulation is a post hoc interpretive lens, offered as a hypothesis to inform future, prospectively validated design work in PE and digital mental health support rather than as a demonstrated result. Full article

Background: Obstructive sleep apnea (OSA) and primary aldosteronism (PA) are linked by bidirectional mechanisms. Although guidelines recommend PA screening in hypertensive patients with OSA, simple tools for clinical prioritization remain limited. We aimed to characterize clinical, polysomnographic, and biochemical features associated with coexisting PA in patients with OSA. Methods: We conducted a cross-sectional study in 107 adults with OSA and definitive PA classification, including 30 patients in the OSA+PA group and 77 in the OSA-only group. Clinical, polysomnographic, and fasting biochemical variables were collected using standardized procedures. Multivariable logistic regression was used to examine factors associated with coexisting PA. Results: Compared with the OSA-only group, the OSA+PA group had a larger neck circumference (42.0 ± 4.6 vs. 39.8 ± 3.7 cm, p = 0.010), a higher snoring index (369.9 ± 320.3 vs. 252.0 ± 264.9, p = 0.040), and higher aldosterone levels (25.8 ± 18.8 vs. 10.7 ± 7.3 ng/dL, p = 0.001), while glycated hemoglobin was lower (6.4 ± 1.6 vs. 7.5 ± 2.1, p = 0.005) and serum potassium was significantly reduced (3.5 ± 0.4 vs. 4.0 ± 0.4 mmol/L, p = 0.001). In multivariable analysis, aldosterone (OR = 1.138, p = 0.001) and serum potassium (OR = 0.017, p = 0.001) were associated with coexisting PA. Conclusions: Higher aldosterone and lower serum potassium were readily available biochemical features associated with coexisting PA among clinically selected patients with OSA. Full article

Objectives: Extracochlear electrodes (EEs), defined as electrode contacts located outside the cochlea, can degrade cochlear implant (CI) performance. We present and validate an Electric Field Imaging (EFI)-based algorithm that helps detect EEs and can be used intra- or post-operatively. Methods: An algorithm for detecting EEs from EFI data was developed. Validation and testing were performed using three datasets: (a) bench models with known EE conditions (saline and resistive load model), (b) clinical EFI recordings from CI recipients with imaging confirmation (CT or plain X-ray), and (c) EFI recordings without imaging. Primary outcomes included the ability to differentiate extracochlear conditions from fully inserted electrode arrays, as well as the concordance between the number of EE contacts identified by the algorithm and those determined from controlled bench configurations or available imaging data. Results: The algorithm reliably differentiated full insertion from EE conditions on bench models and clinical EFI recordings from CI recipients with imaging confirmation. In the imaging-confirmed clinical cohort (6 EE cases among 226 CI recipients), the algorithm achieved 100% sensitivity and specificity. In clinical cases without imaging, the algorithm flagged 3.94% as having extracochlear electrodes, within the range of prevalence reported in the literature. Furthermore, the lateral-wall electrode arrays were associated with a significantly higher extracochlear electrode occurrence compared to the pre-curved arrays. Conclusions: This EFI-based algorithm may provide a useful screening tool for EEs using routine clinical measurements, supporting intra-operative and post-operative detection while reducing reliance on imaging and thereby aiding clinical decision-making. The method can be integrated into existing clinical fitting software workflow and complements recent work on EFI-based tip fold-over detection. The observed sensitivity and specificity suggest that this approach may provide a valuable tool for identifying EEs in situations where imaging is unavailable or impractical. Full article

Objective: Obstructive sleep apnea (OSA) during pregnancy is associated with adverse maternal and fetal outcomes, yet commonly used screening questionnaires may perform differently because of pregnancy-specific physiological changes. This study compared questionnaire-based OSA risk classifications obtained using the Berlin Questionnaire and STOP-BANG and examined their associations with metabolic parameters and diabetes-related history. Methods: This cross-sectional study included 210 pregnant women attending a tertiary obstetrics clinic. Berlin Questionnaire high risk was defined as positivity in at least two categories. STOP-BANG scores of 0–2, 3–4, and 5–8 indicated low, moderate, and high risk, respectively. Sociodemographic, obstetric, anthropometric, and biochemical variables were analyzed using multivariable logistic regression. Results: The Berlin Questionnaire classified 34 participants (16.2%) as high risk. STOP-BANG classified 183 (87.1%) as low risk, 5 (2.4%) as moderate risk, and 22 (10.5%) as high risk. Higher current weight was associated with increased questionnaire-based OSA risk according to both tools. In the Berlin model, HbA1c, BMI, and family history of diabetes were independently associated with high questionnaire-based OSA risk. No independent associations were identified in the STOP-BANG model. Although the Berlin model had 83.3% overall accuracy, it correctly classified only 2.9% of high-risk participants. Conclusions: The questionnaires identified different patterns of OSA risk. Findings were instrument-specific and do not indicate confirmed OSA. Given the absence of polysomnography, modest model performance, and limited high-risk classification, the results should be interpreted cautiously. Full article

Age-related sarcopenia involves structural and functional neuromuscular changes. Electrophysiological measures from surface electromyography (sEMG) capture activation dynamics, spectral fatigue indices and motor unit properties that may constitute objective signatures of sarcopenia. The objective of this work is to systematically review sEMG features, fatigability metrics and AI-based classification/regression approaches reported for sarcopenia assessment between 2019 and 2026. PRISMA guidelines were followed, and IEEE Xplore, PubMed and Scopus were searched for open access human studies. Extracted information comprised sample characteristics, muscles and tasks, signal acquisition and preprocessing, extracted time/frequency/time–frequency and motor unit features, fatigue metrics, machine learning pipelines, validation schemes and dataset accessibility. Studies were classified into activation, fatigue, ML, and neural control groups; risk of bias was assessed. A total of 12 studies fulfilled the inclusion criteria. Recurrent electrophysiological signatures included reduced distal activation with compensatory proximal recruitment and higher antagonist co-activation; diminished MF/IMDF fatigue slopes indicative of Type II fiber loss and altered motor unit recruitment; motor unit analyses revealed decreased discharge rates and larger MUAP amplitudes. AI-based models combining multidomain features (time, spectral, CWT/EMD, and motor unit metrics) yielded reasonable screening performance (AUC/accuracy 0.73–0.89) when using robust feature selection and explainability tools. Heterogeneity in acquisition, normalization, small cohorts and sparse data sharing limited comparability and external validity. The findings indicate that sEMG-derived electrophysiological signatures are promising for sarcopenia detection and monitoring. To translate signatures into reliable clinical tools, standardized protocols, larger shared datasets, multimodal features, including motor unit metrics, and rigorous external validation of AI models are required. Full article

Objectives: This systematic review synthesized the existing literature to evaluate the associations between maternal Selenium (Se) exposure during pregnancy and neonatal outcomes, specifically focusing on parameters of mortality, morbidity, and neurobehavioral development. Methods: PubMed, Embase, and the Cochrane Library were systematically searched from inception through December 2025. For quality appraisal, Cochrane RoB2 and the National Heart, Lung, and Blood Institute of the National Institutes of Health tools for observational studies were applied. Results were qualitatively synthesized. Results: Screening of 2743 unique records and 473 full-text articles yielded 355 papers included in the overarching SeduP project, with 19 specific to this review. Synthesized evidence revealed a lack of consistent evidence of an association between maternal Se status and all-cause neonatal or perinatal mortality (n = 413, 4 studies), combined mortality/morbidity (n = 2753, 2 studies), or infectious-related morbidity (n = 928, 3 studies). Conversely, conflicting findings were observed regarding all-cause neonatal morbidity (n = 68, 2 studies) and neurobehavioral development (n = 1442, 4 studies), while a potentially inverse association was identified exclusively between Se levels and pulmonary-related neonatal morbidity (n = 611, 4 studies). Conclusions: Despite compelling preclinical biological plausibility, clinical data do not support a widespread protective effect of maternal Se status against adverse neonatal outcomes. The existing body of evidence is heavily constrained by moderate to very low quality, driven by small sample sizes, insufficient confounding adjustments, and suboptimal exposure assessment timing at delivery. While localized signals imply potential pulmonary protection alongside risks of excess toxicity, large-scale, standardized prospective cohorts are critically needed to generate clinically actionable data. Full article

Artificial intelligence (AI) is increasingly being applied in healthcare, with growing relevance to clinical nutrition. This narrative review examines current and emerging uses of AI in nutrition care within the Nutrition Care Process framework, with attention to assessment, monitoring and evaluation, diagnosis, intervention, and clinical support tools. Current applications include AI-assisted dietary assessment using image recognition, wearable sensors, analysis of continuous glucose and other physiologic data for early risk detection, and support for malnutrition screening and diagnosis. AI is also being explored for identifying micronutrient deficiencies and complications of nutrient excess, as well as for screening and early intervention in eating disorders. In nutrition intervention, AI has potential to support personalized dietary planning, nutrition support in intensive care settings, behavioral interventions, and precision nutrition approaches such as digital twins. Additional applications include clinical decision support and documentation assistance. However, despite its usefulness, concerns about AI systems exist. Its performance depends on the quality of the data used to train it; it can introduce bias, and it can produce inaccurate or misleading outputs. In addition, overreliance on AI may also reduce clinician attentiveness and contribute to cognitive errors. For these reasons, AI should be regarded as a support tool rather than a replacement for human clinical care. Overall, AI offers substantial opportunities to improve the personalization, efficiency, and scalability of clinical nutrition practice, but its safe and effective implementation will require continued validation, careful oversight, and integration with clinical expertise. Full article

Background/Objectives: Takotsubo cardiomyopathy (TTC), also known as “broken heart syndrome”, is an acute and reversible cardiac condition often triggered by emotional or physical stress. Increasing evidence suggests a strong comorbidity between TTC and psychiatric disorders, particularly anxiety and depression. However, the nature, prevalence, and implications of these associations remain incompletely understood. This scoping review aimed to systematically map the available evidence on the association between Takotsubo cardiomyopathy and anxiety and depressive disorders, characterize the prevalence and nature of these psychiatric comorbidities, evaluate the methods used for their assessment, and identify knowledge gaps to guide future research and clinical practice. Methods: A systematic search of MEDLINE (via PubMed), Embase, Web of Science, and Google Scholar was conducted for peer-reviewed human studies published up to December 2025 in English or French. Eligible studies reported on psychiatric comorbidities in TTC populations, including pre-existing disorders, psychiatric triggers, or psychological sequelae. A standardized extraction grid and the Joanna Briggs Institute appraisal tools were used for data collection and quality assessment. Results: Eighteen studies were included. Depression and anxiety were prevalent among TTC patients, frequently preceding or accompanying cardiac events. Emotional stressors such as bereavement or panic attacks were commonly cited as triggers. Despite this, only a minority of studies used validated psychometric tools, and longitudinal psychiatric follow-up was rare. Most studies were limited by small sample sizes, retrospective designs, and heterogeneity in diagnostic approaches. Conclusions: The findings reveal a consistent association between TTC and affective disorders, underscoring the role of the brain–heart axis. Integrating psychiatric screening and follow-up into TTC management may enhance outcomes. Future interdisciplinary research is needed to clarify causal pathways and inform prevention and treatment strategies. Full article

Background/Objectives: Preeclampsia is a heterogeneous hypertensive disorder of pregnancy, and accessible hematologic indices may reflect maternal physiologic differences before the clinical syndrome becomes evident. This study evaluated whether the first-trimester hemoglobin-to-red cell distribution width (Hb/RDW) ratio was associated with later documented preeclampsia and assessed its discrimination, relation to disease severity, and association with maternal–neonatal outcomes in a tertiary-care cohort. Methods: This retrospective cohort study included 224 singleton pregnancies managed at a tertiary referral center between January 2024 and December 2025. Data extraction, anonymization, and analysis were performed after ethics approval (Approval No. 72; 2 March 2026). Participants were categorized as preeclampsia-negative (n = 131) or preeclampsia-positive (n = 93). Preeclampsia and severe features were defined according to American College of Obstetricians and Gynecologists criteria. The Hb/RDW ratio was calculated from routine first-trimester complete blood count parameters obtained before clinical diagnosis or final outcome classification. ROC analysis, multivariable logistic regression, multicollinearity assessment, events-per-variable evaluation, calibration testing, and exploratory incremental discrimination analyses were performed. Results: The first-trimester Hb/RDW ratio was higher in pregnancies later complicated by preeclampsia. Single-marker ROC analysis showed moderate discrimination (AUC = 0.687; bootstrap 95% CI: 0.618–0.755). At the 0.63 cut-off, sensitivity was 89.2%, specificity 47.3%, positive predictive value 54.6%, and negative predictive value 86.1%. In the primary multivariable model, Hb/RDW remained associated with later preeclampsia (adjusted OR per 0.1-unit increase = 1.45; 95% CI: 1.22–1.72; p < 0.001). However, Hb/RDW was not independently associated with severe preeclampsia among affected patients, fetal growth restriction, or composite maternal morbidity. VIF values ranged from 1.00 to 1.07. Secondary models for fetal growth restriction and composite maternal morbidity had borderline events-per-variable values and were interpreted as exploratory. Conclusions: First-trimester Hb/RDW was associated with later documented preeclampsia in this retrospective tertiary-care cohort, but its single-marker performance was modest and specificity was limited. The ratio should be interpreted as an inexpensive research marker of hematologic phenotype rather than a stand-alone screening, diagnostic, or management tool. Prospective multicenter validation with standardized first-trimester sampling and adjustment for hematinic and nutritional variables is required before clinical implementation. Full article

Whole-genome sequencing has become a central approach for investigating candidate probiotic bacterial isolates, but probiogenomic analysis often requires multiple independent tools, manual marker-gene searches and study-specific interpretation of heterogeneous outputs. PROBEAT (PRObiotic Bacterial gEnome Analysis Toolkit), a Dockerized workflow for the integrated and reproducible genome-based characterization of candidate probiotic isolates. PROBEAT combines standard bacterial genome analysis with specialized probiotic-oriented features, covering read processing, genome assembly and quality assessment, taxonomic confirmation, genome annotation, safety screening, functional and metabolic profiling, probiotic marker detection and automated visualization. A key feature of PROBEAT is a custom Probiotic Gene Markers database containing more than 400 gene aliases organized into 320 curated marker records and 32 functional categories. The workflow automatically converts raw sequencing data into a structured PDF booklet report that integrates safety, taxonomy, metabolism, functional annotation and probiotic-oriented data. PROBEAT reduces the technical burden of bacterial genome analysis and provides a reproducible framework for standardized genome-based assessment of probiotic potential, while supporting functional annotation rather than clinical claims. Full article

Background/Objectives: Psoriatic arthritis (PsA) affects up to 30% of psoriasis patients but remains underdiagnosed due to heterogeneous early manifestations. Existing tools rely on advanced imaging or biomarkers, limiting feasibility. This study aimed to develop a simple nomogram, using routinely available clinical features to identify patients with likely PsA. Methods: This multicenter cross-sectional study enrolled adult psoriasis patients from two tertiary hospitals (training: n = 884; validation: n = 690). PsA was diagnosed using the Classification Criteria for Psoriatic Arthritis (CASPAR) by rheumatologists and dermatologists who were blinded to each other’s assessments. Thirteen candidate predictors, including demographics, disease history, and lesion characteristics, were prespecified. Independent predictors were selected using least absolute shrinkage and selection operator (LASSO) regression, followed by multivariable logistic regression. Internal validation used 1000 bootstrap resamples, followed by temporal–geographic validation. Performance was assessed using area under the curve (AUC), calibration plots, the Hosmer–Lemeshow test, decision curve analysis (DCA), and clinical impact curves (CIC). Sensitivity analyses included E-value analysis for unmeasured confounding and alternative model specifications. Results: Five independent predictors were identified: uveitis (odds ratio (OR): 5.23, 95% confidence interval (CI): 1.94–14.10), inverse lesions (OR: 1.65, 95% CI: 1.09–2.48), scalp lesions (OR: 2.97, 95% CI: 1.45–6.10), nail lesions (OR: 5.98, 95% CI: 3.69–9.68), and arthralgia (OR: 20.23, 95% CI: 13.14–31.13). The nomogram showed good discrimination (AUC: 0.885 training, 0.874 validation), with good calibration and clinical utility confirmed by DCA and CIC. Conclusions: This nomogram, integrating five clinically accessible predictors, offers a practical tool to identify psoriasis patients with likely PsA, supporting timely referral in tertiary dermatology outpatient settings. Full article

Objectives: The purpose of this study was to evaluate the ability of computer vision models to detect myopia from standard colour fundus photographs and to compare their diagnostic performance with that of experienced ophthalmologists. Methods: A previously published dataset of 324 retinal fundus images labelled as myopic or non-myopic based on cycloplegic refraction as used for model training and internal validation. Images were acquired using a non-mydriatic 45° fundus camera. Final model evaluation was performed on an independent test set of 50 images from different patients who were not included in the original dataset. YOLOv8 and YOLOv11 variants were trained for binary classification. Internal validation used patient-level cluster bootstrap confidence intervals, whereas image-level bootstrap confidence intervals were estimated for the independent test set. Pairwise model comparisons were adjusted using the Holm–Bonferroni correction. Five experienced ophthalmologists independently classified the test set, and their consensus was compared with the selected YOLO models using DeLong’s and exact McNemar tests. Results: Internal validation identified YOLOv8-m and YOLOv11-n as the best-performing models according to a predefined composite score used exclusively for model selection. On the independent test set, YOLOv11-n achieved the highest area under the curve (AUC = 0.889), followed by YOLOv8-m (0.806), although the difference was not statistically significant (DeLong test, p > 0.05). The clinical consensus achieved an AUC of 0.832, with no significant difference compared with either model. Exact McNemar testing likewise revealed no statistically significant differences in paired classification outcomes between either AI model and the clinical consensus. Limitations include the small, single-centre, class- and age-imbalanced dataset and the limited number of expert observers. Conclusions: Although neither YOLOv8-m nor YOLOv11-n showed statistically significant differences from the clinical consensus on this independent test set, these findings should be interpreted cautiously given the relatively small, single-centre study population. Larger multicentre studies with independent external validation are warranted to confirm the generalisability, robustness, and potential role of clinician-driven computer vision models as decision support tools for myopia screening. Full article