搜索结果：Class

Convolutional neural networks (CNN) for skin cancer classification have shown results comparable to dermatologists but are vulnerable to minor image transformations. We investigated the robustness of a MDR class-IIa certified CNN when classifying sequential images of identical lesions. We acquired 2,744 dermoscopic images of 385 skin lesions (80.8 % benign, 19.2 % malignant) and applied in-vivo zoom, rotation (90-degree increments), and simple repetitions of image recordings. Sequential images of identical lesions were classified by a binary CNN (Moleanalyzer-Pro, FotoFinder Systems, Germany) and the variability of scores was investigated using intraclass correlation coefficient (ICC), mean absolute change of scores (mac), and probability of change of predicted class ( π c l a s s c h a n g e ${{\pi }_{class\ change}}$ ). In dermoscopic baseline images (n = 385) the CNN showed a sensitivity, specificity, and area under the receiver operating characteristic (AUROC) (95 % CI) of 91.9 % (83.4 %-96.2 %), 87.8 % (83.7 %-91.0 %) and 0.947 (0.921-0.972), respectively. The ICC across images of identical lesions was 0.872 (0.862-0.883), indicating excellent reliability. Overall mac of scores was 0.102 (0.090-0.115) and π c l a s s c h a n g e ${{\pi }_{class\ change}}$ was 7.5 % (5.8 %-9.2 %). The tested CNN demonstrated a profound robustness against image variations as might be introduced during sequential digital dermoscopy. Clinically relevant class changes occurred in one of 13 images.

While initial anhedonia predicts poor psychotherapy outcomes, little is known about its trajectory during treatment. This study aimed to: (1) identify distinct anhedonia trajectories during high-intensity depression treatment; (2) examine patient and treatment predictors; and (3) compare outcomes across treatment types. Sessional anhedonia scores (PHQ-9 item-1) from 22,605 patients in NHS talking therapies (primarily receiving either cognitive-behavioral therapy [CBT] or counseling for depression [CfD]) were analyzed using latent growth curve (LGC) and growth mixture modeling. Multinomial logistic regression examined predictors of class membership. A quadratic LGC model best fit the data, reflecting a decrease in symptoms before leveling out. Six latent classes emerged. Notably, three "non-responder" classes characterized by linear-stable or minimal-change patterns comprised over 50% of the sample (51.3%). In contrast, two "responder" classes (41.4%) exhibited improvement, typically shifting between sessions 4 and 6. This suggests an early "inflection point" where the trajectory of recovery is established. Poorer response was predicted by unemployment, chronic health conditions, psychotropic medication, and longer wait times. There was only a sufficient sample size to compare CBT and CfD treatment types. While CBT was associated with membership in specific classes, the probability of being a "responder" did not differ significantly between CBT and CfD. Most patients followed non-responder trajectories, highlighting a major efficacy gap for anhedonia in standard depression protocols. The 4-6 session window suggests that if improvement is not observed early, the treatment strategy may require further evaluation. Further research into targeted anhedonia interventions is essential.

Our study aimed to quantify the extent to which individuals, when considering a hypothetical abortion at eight weeks gestation, accept tradeoffs across attributes of abortion care options. We also sought to evaluate the presence of different preference patterns and whether they are associated with respondent characteristics. We designed a discrete-choice experiment based on qualitative interviews with providers, abortion advocates, and individuals with personal experience seeking abortion. Each respondent saw eight choice questions with two experimentally designed abortion care options described by study attributes of abortion type, wait time, travel distance, electronic documentation, and out-of-pocket cost and a no-abortion option. Respondents were recruited across abortion clinics, websites and a commercial panel. We estimated respondents' choices using random-parameters logit and latent-class analysis. Overall, 500 individuals completed the survey. The majority (80.2%) of respondents chose one of the abortion care options across all choice questions; 18.2% chose the no-abortion option at least once. Latent-class analysis revealed three preference patterns: one class representing individuals favoring any abortion type that avoided a 4-week wait time and without electronic documentation and another representing individuals with choices highly sensitive to out-of-pocket costs. A small third class provided no preference information. Respondent characteristics were not associated with latent-class membership. While out-of-pocket costs are a primary concern for some abortion seekers, a sizeable subset prioritizes expediency and privacy. Recognizing this preference heterogeneity may help abortion clinicians and support organizations deliver more patient-centered care. Relevant to current threats to mifepristone availability, participants would still choose abortion even if their preferred type and mode of delivery were not available. Multi-pronged strategies to decrease wait times and to reduce out-of-pocket costs would help meet the needs and priorities of abortion seekers.

Engagement-students' active involvement in learning activities-enables achievement and is enabled by interpersonal support. In a physical education class, one's teacher and one's classmates potentially provide such interpersonal support. The present study focused specifically on agentic engagement (the student's constructive input into the learning environment) and on peer support (one's classmates' effort encouragement, improvement focus, and relatedness support) to propose and test a reciprocal effects model in which agentic engagement and peer support positively facilitated each other over time. In a preregistered, three-wave, within-person survey study, 740 South Korean secondary students (41.6% females; Mage = 15.4) from 24 physical education classes completed the same questionnaire at the beginning, middle, and end of an 18-week semester. We tested for reciprocal effects, using a random-intercept cross-lagged panel model. Peer-directed agentic engagement prospectively predicted subsequent perceived supportive peer climate (B = 0.32, SE = 0.10, t = 3.29, p = .001), and perceived supportive peer climate prospectively predicted subsequent peer-directed agentic engagement (B = 0.20, SE = 0.06, t = 3.51, p < .001). These findings suggest a mutually facilitative process in physical education in which a student's agentic contributions into their peer interactions and their perceptions of peer support during those interactions prospectively, reciprocally, and positively facilitate each other over time.

We evaluated the association between pre- and/or postoperative heart failure guideline-directed medication (GDMT) prescribing on mid-term survival following CABG. We included the records of 4,307 adult patients undergoing outpatient, elective isolated CABG from 1/2016-12/2024 discharged alive, and merged our institutional STS database with electronic health record medication data. GDMT class prescribing pre- (within 2 weeks of admission) and at discharge was identified: beta blockers, angiotensin converting enzyme inhibitors/receptor blockers/neprilysin inhibitors (ACE/ARB/ARNI), sodium-glucose cotransporter 2 inhibitors (SGLT2), and mineralocorticoid receptor antagonists. The primary outcome was mortality on follow-up. 4,307 elective, isolated CABG patients were included with mean age 64&#xb1;11 years and mean STS PROM score 1.3&#xb1;1.4%. Preoperatively, 5.2% (222/4,307) of all patients and 13.3% (40/300) of patients with an EF&#x2264;40% & GFR>30 mL/min were admitted on 3+ GDMT medication classes. Postoperatively, 25.0% (1,078/4,307) of all patients and 56.3% (169/300) of EF&#x2264;40% & GFR>30 mL/min patients were discharged on 3+ classes. Over a median of 426 days of follow-up (IQR 34-1190), 7.2% (309/4,307) of patients died. One- and three-year overall survival was 96.2% and 90.9%. Though no protective association was seen for preoperative GDMT, an increasing number of postoperative GDMT medications was independently associated with reduced follow-up mortality (all patients: HR 0.64, 0.55-0.74, p<0.001), even after controlling for postoperative complications. The strongest independent associations were noted for post-operative SGLT2 (HR 0.39, 0.27-0.56, p<0.001) and ACE/ARB/ARNI (HR 0.61, 0.45-0.83, p=0.001). The number of postoperative GDMT medication classes prescribed at discharge following CABG was strongly associated with reduced mortality.

Real-world data on calcitonin gene-related peptide (CGRP) monoclonal antibodies (mAbs) are essential to determine whether a class effect exists and to assess the benefit of switching between treatments. Available data remain limited as most studies focused exclusively on patients who failed previous CGRP mAbs and evidence regarding newer agents such as eptinezumab are still scarce. This prospective real-world study included all adult patients enrolled in the FHU InovPain registry who received intravenous eptinezumab 100 mg and after prior treatment with one or more subcutaneous CGRP mAbs, regardless of their response to previous CGRP mAbs. According to the 50% response rate (in terms of monthly migraine days) after 6 months of treatment, a descriptive analysis of switching from subcutaneous CGRP mAbs to eptinezumab was performed. Patients were classified into three subgroups: cross-effectiveness (all CGRP mAbs used were effective), cross-ineffectiveness (all CGRP mAbs used were ineffective), and no cross-response (different responses across CGRP mAbs used). Factors associated with response to CGRP mAbs were investigated by comparing patients with cross-ineffectiveness (with the CGRP mAbs used) to those who responded to at least one CGRP mAb (cross-effectiveness and no cross-response), followed by multivariate logistic regression. A total of 190 patients (83.7% women; mean age 52.2 &#xb1; 13.7 years) were included. The 50% responder rate to eptinezumab was 76.0% (95% CI: 67.3-83.1) in patients who had responded to at least one previously used CGRP mAb, compared with 30.4% (95% CI: 20.2-42.8) in patients with no prior response to CGRP mAbs. Cross-effectiveness, cross-ineffectiveness, and no cross-response were observed in 46.8% (95% CI: 39.6-54.2), 28.9%, (95% CI: 22.7-36.0) and 24.2% (95% CI: 18.4-31.7) of patients, respectively. Only two factors were associated with response to at least one CGRP mAb: a lower helplessness score on the Pain Catastrophizing Scale (AdjOR 0.91, 95% CI: 0.86-0.97, p=0.004) and a lower allodynia score on the ASC-12 (AdjOR 0.91, 95% CI: 0.84-0.98, p=0.010). This real-world study confirms the clinical benefit of switching to eptinezumab in nearly one-third of patients who did not respond to previous subcutaneous CGRP mAbs. It also demonstrates a class effect that is not absolute, as nearly one-quarter of patients showed no cross-response between CGRP mAbs. Not applicable.

To conduct a preliminary single-center feasibility study of a YOLO-based deep-learning model for automated detection of lumbar disc herniation at the L4-L5 and L5-S1 levels on sagittal MRI. In this retrospective study, 372 anonymized sagittal T2-weighted lumbar MRI slices from adult patients evaluated for low back pain at a single tertiary center were reviewed. Intervertebral discs at the L4-L5 and L5-S1 levels were labeled with bounding boxes into four classes (L4-L5 herniated, L5-S1 herniated, L4-L5 non-herniated, L5-S1 non-herniated) by two clinicians (a radiologist and a PM&R specialist, each with 6 years of post-residency experience), with consensus adjudication of disagreements, using the combined North American Spine Society / ASSR / ASNR nomenclature as reference. The dataset was split into 337 training, 25 validation, and 10 test images. Rotation-based augmentation was applied. A YOLO object-detection architecture was trained and evaluated using precision, recall, F1, and mean average precision (mAP@0.5 and mAP@0.5-0.95). 95% bootstrap confidence intervals were estimated for aggregate metrics. The model achieved an overall precision of 0.738, recall of 0.698, mAP@0.5 of 0.744, and mAP@0.5-0.95 of 0.454, with a best overall F1 of 0.69 at a confidence threshold of approximately 0.30. Herniated classes outperformed non-herniated classes, with the highest recall observed for L5-S1 herniated discs (0.963). Bootstrap confidence intervals were wide, consistent with the small test set. This preliminary feasibility study suggests that YOLO-based level-specific detection of lumbar disc herniation at L4-L5 and L5-S1 on sagittal MRI is technically feasible. Given the very small test set, single-center design, absence of external validation, and lack of radiologist benchmarking, these results are hypothesis-generating and not yet sufficient to support any clinical or autonomous-diagnostic use. Larger multi-center validation and prospective comparison with radiologists are required.

Serum sialic acid binding immunoglobulin-like lectin-1 (sSIGLEC-1) is a type I interferon-associated biomarker previously linked to lupus nephritis (LN) in European ancestry populations, but its utility in non-European cohorts remains poorly defined. This study aimed to validate the association of sSIGLEC-1 with LN in Egyptian SLE patients (non-European ancestry) and to test its correlation with world health organization (WHO) pathological classes, chronic kidney disease (CKD) stages, systemic lupus international collaborating clinics-renal activity score (SLICC-RAS), systemic lupus erythematosus disease activity index (SLEDAI), 24-hour urinary protein and proinflammatory cytokines. This cross-sectional study included 80 SLE patients (47 with LN, 33 without LN) and 20 healthy controls. Renal biopsy was classified according to WHO criteria. Estimated glomerular filtration rate (eGFR) and CKD stages were calculated. Median levels of sSIGLEC-1 were significantly higher in SLE patients than controls (113.5 vs. 11.2 pg/mL) and in LN patients than non-LN patients (117.7 vs. 110.0 pg/mL). Multivariable logistic regression confirmed sSIGLEC-1 as an independent predictor of LN (OR&#x2009;=&#x2009;1.02, p&#x2009;=&#x2009;0.04). sSIGLEC-1 correlated positively with SLEDAI (r&#x2009;=&#x2009;0.26), SLICC-RAS (r&#x2009;=&#x2009;0.31), and proinflammatory cytokines (IL-1&#x3b2;, IL-6, TNF-&#x3b1;), but did not correlate with 24-hour urinary protein (r = -&#x2009;0.175). However, no significant differences in sSIGLEC-1 were observed across WHO pathological classes or CKD stages. ROC analysis showed poor discriminative ability for LN (AUC&#x2009;=&#x2009;0.6928, sensitivity 93.6%, specificity 39.4%). In Egyptian SLE patients, sSIGLEC-1 is elevated in LN and correlates with disease activity but does not reflect histological severity or chronic kidney damage. Its low specificity limits diagnostic utility; however, high sensitivity suggests potential as a rule-out screening test for LN in non-European populations.

Neoantigens-tumor-specific peptides generated by somatic mutations-are central targets of effective anticancer T cell immunity and underpin the clinical success of immune checkpoint blockade and personalized cancer vaccines. Advances in high-throughput sequencing, immunopeptidomics, and artificial intelligence (AI) have transformed neoantigen discovery from tailored experimental workflows into scalable, computational pipelines. However, accurately identifying the small subset of tumor mutations that yield processed, presented, and immunogenic epitopes remains a major bottleneck. This review summarizes how AI is reshaping neoantigen discovery, from somatic variant calling, HLA typing, and peptide processing to peptide-MHC binding, presentation, and T cell recognition. We first outline the immunobiological foundations of antigen presentation, emphasizing class I and II peptide-binding grooves and their allele-specific motifs, then describe AI workflows that integrate somatic mutation calling, HLA typing, transcriptomics, and immunopeptidomics to nominate candidate neoepitopes. We highlight recent AI-driven tools for presentation and immunogenicity prediction, integrative pipelines that support personal and shared neoantigen targeting, and early clinical applications in vaccination and T cell therapies. AI-driven models trained on eluted ligand datasets substantially outperform affinity-only predictors for peptide presentation across diverse HLA alleles and populations. Consortium-scale benchmarking demonstrates that integrating features of antigen processing, presentation, and TCR recognition can eliminate the majority of non-immunogenic candidates while retaining clinically relevant neoepitopes. Immunopeptidomics provides essential ground truth, revealing that only a small fraction of genomically predicted candidates are naturally presented and uncovering noncanonical antigen sources, including splice variants, post-translational modifications, and noncoding regions. Integrative pipelines now support both personal (private) and shared (public) neoantigen prioritization, enabling translational applications such as personalized vaccines and TCR-based therapies. AI-guided neoantigen discovery is now clinically actionable, enabled by immunopeptidomics and deep learning models. Despite significant progress, key challenges remain, including limited class II prediction accuracy, incomplete coverage of rare HLA alleles, tumor heterogeneity, and the need for standardized benchmarking and validation. Anchoring computational predictions to mass spectrometry-derived ligands and incorporating tumor evolution and immune escape mechanisms will be critical for improving target selection. Continued integration of AI, proteogenomics, and clinical data is poised to accelerate the development of effective, precision neoantigen-based cancer immunotherapies.

In the present study, the chemical compositions of the essential oils from aerial parts of Centaurea heywoodiana Raimondo, Spadaro & Di Gristina (sect. Dissectae (Hayek) Dostal.) and Centaurea phalacrica Brullo, Cambria, Crisafulli, Tavilla & Sciandr. (sect. Aplolepidae (J. Arenes) Dostal.), two completely unexplored endemic species of Sicily, were evaluated by GC-MS. The main components of the essential oil of C. heywoodiana (Ch) were sesquiterpenes (52.2%), with caryophyllene oxide (29.5%) as the most abundant constituent of the oil. Carbonylic compounds (43.8%) were the principal class of the essential oil of C. phalacrica (Cp) being hexanal (12.1%) as the main metabolite. Sesquiterpene hydrocarbons were also well represented (35.4%), with &#x3b2;-elemene (11.7%) as the main metabolite of the class. Furthermore, a complete literature review on the composition of the essential oils of all the other taxa of Centaurea, belonging to sections Dissectae and Aplolepidae, studied so far, was carried out.

Fraxinus mandshurica, a deciduous tree species in the Oleaceae family, is valued for its high-quality timber, medicinal properties, and ecological functions. It is classified as a second-class national protected wild plant in China. However, the complete sequence information of its mitochondrial genome has yet to be reported, and systematic studies on its genetic background and phylogenetic evolution remain limited. The mitochondrial genome of F. mandshurica was assembled into a linear structure, measuring 645,536 bp and 44.40% GC content. Annotation of the mitochondrial genome yielded 77 genes, including 44 protein-coding genes, 30 tRNAs, and 3 rRNAs. A total of 585 dispersed repeats, 184 simple repeats, and 21 tandem repeats were detected, and 476 RNA editing sites with C-to-U type changes were predicted. Codon usage analysis revealed a bias toward codons terminating with A/T. Furthermore, homology analysis revealed 36 chloroplast-derived fragments, spanning 55,673 bp and accounting for 8.62% of the mitochondrial genome. Collinearity analysis exposed considerable genomic rearrangements between F. mandshurica and its close relatives, reflecting structural divergence across evolution. Positive selection was observed for 14 genes, in which Ka/Ks ratios exceeded 1 in at least one pairwise comparison. Analysis of nucleotide polymorphisms showed differences in every gene, where nad4 displayed the greatest variation. Phylogenetic analyses suggested that F. mandshurica and F. excelsior are closely related. This study presents an integrated analysis of the F. mandshurica mitochondrial genome, revealing its unique structural and evolutionary characteristics. Moreover, this mitochondrial genome was compared with seven published Oleaceae mitochondrial genomes to investigate genome dynamics across the family. Our findings enrich the mitochondrial genome resources for Oleaceae species and highlight the potential of mitochondrial genes to elucidate plant evolutionary history.

Postoperative delirium (POD) is a prevalent complication among older surgical patients; however, accurately predicting its occurrence using traditional point-based risk scores remains challenging. This study aimed to develop a delirium prediction model using a contemporary dataset and to compare the performance of modern machine learning (ML) algorithms with that of the Inouye and Marcantonio risk indices. A retrospective cohort of 283 surgical patients aged &#x2265; 50 years (POD incidence = 8.8%) was analyzed. Candidate predictors included demographic characteristics, frailty markers, comorbidities, surgical category, and planned intensive care unit (ICU) admission. Class imbalance was addressed using the Synthetic Minority Over-sampling Technique. Five classifiers-ridge logistic regression (baseline), random forest, XGBoost, CatBoost, and a stacking ensemblewere trained using 5-fold stratified cross-validation. Model discrimination, assessed by the area under the receiver operating characteristic curve (AUC), along with F1 score, precision, and recall, was averaged across folds. For comparison, the Inouye 4-factor and Marcantonio 7-factor scoring systems were recalculated for each patient. The mean patient age was 65.7 &#xb1; 8.6 years, and 40.6% were female. The incidence of POD varied significantly across surgical specialties, ranging from 18.4% in hepatobiliary procedures to 3.7% in general and vascular surgeries. Univariate analysis identified older age, functional dependency, planned ICU admission, and high-risk surgery as significant risk factors. The AUCs of the ML models ranged from 0.57 (logistic regression) to 0.78 (stacking ensemble). Among individual models, CatBoost achieved the best overall balance (AUC = 0.77). In contrast, traditional risk scores demonstrated poor performance (Inouye AUC &#x2248; 0.50; Marcantonio &#x2248; 0.56). Contemporary ML models outperformed conventional delirium risk scores in predicting POD, although their overall discrimination remained moderate. Key predictors included age, frailty markers, and surgical complexity. Integration of ML-based screening into electronic health records may enable earlier identification of high-risk patients and facilitate targeted delirium prevention strategies. External validation in larger, multi-center cohorts is warranted to confirm these findings.

Dysphagia is a common complaint that afflicts the elderly and individuals with neurological disorders. Conventional diagnostic techniques, such as Videofluoroscopic Swallowing Studies (VFSS) and Fiberoptic Endoscopic Evaluation of Swallowing (FEES), have disadvantages including inter-rater variability, radiation exposure, invasive procedures, the need for clinician observation, and subjectivity. The review aims to highlight applications of Artificial Intelligence (AI) using Deep Learning (DL) and machine-learning methods to diagnose dysphagia, with the objective of increasing diagnostic accuracy, objectivity, and specificity. A literature review was performed using the following databases: PubMed, Scopus, Web of Science, and Google Scholar, covering the period from January 2011 to December 2025. These articles were selected based on the use of AI for dysphagia Evaluation and the reporting of diagnostic metrics. The AI models, such as wearable sensor technologies and Inflated 3D Convolutional Neural Networks, achieved higher classification accuracy of 95.96% and success detection of 97.5% for the swallowing reflex class than conventional diagnostic modalities. The other side of AI model incorporation attenuates the role of traditional diagnostic methods by enabling continuous monitoring of objective device data. However, even among more technologically advanced methods, the constraints and complexity of AI applications pose a barrier to calls for sex-specific model calibration. Artificial Intelligence represents a dramatic advancement in dysphagia diagnosis, offering greater accuracy, diagnostic objectivity and real-time analysis. Nonetheless, realising their full potential requires coordinated efforts to incorp&#xf3;rate these computational techniques into standard clinical bedside practice.