Browse Articles

Patient address parsing via KG-aware contrastive learning and constrained on-prem LLM inference

Abstract Address parsing seeks to map noisy, abbreviated free-text addresses into standardized hierarchical tuples for large-scale information systems. Existing approaches struggle with semantic and structural ambiguity, hallucination from unconstrained generation, and deployment constraints under privacy and governance requirements. We present AddrKG-LLM, a two-stage framework that combines knowledge-graph (KG)–aware retrieval with schema-restricted large language model (LLM) decoding. First, contrastive learning over multi-view administrative graphs yields node embeddings that retrieve and re-rank a compact Top-K candidate set, thereby bounding the search space while preserving high gold coverage (Recall@K). Second, a candidate-restricted decoder running on-premises produces JSON-compliant outputs, enforcing single-candidate field consistency and alignment with KG priors to improve controllability and policy compliance. Using de-identified real-world records, we evaluate structural consistency via micro-level accuracy ( $$A_{micro}$$ ) and macro-level accuracy ( $$A_{macro}$$ ), and assess system properties with Recall@K and latency. Across strong string-matching, sequence-labeling, and generic LLM baselines, AddrKG-LLM delivers consistent gains in $$A_{micro}$$ and $$A_{macro}$$ with a favorable Recall@K. The proposed method consists of three components: (i) multi-view graph aggregation, (ii) a hierarchy-aware self-supervised contrastive objective that derives positives/negatives from administrative relations to align textual and graph embeddings, and (iii) candidate-restricted decoding within the KG-derived Top-K set. Overall, coupling KG-aware retrieval with constrained on-prem LLM decoding yields an accurate, controllable, and deployable solution for noisy-address structuring across domains.

Patients’ and relatives’ perspectives on the quality of end-of-Life care in advanced cancer: From the final months to bereavement

Background End-of-life care affects both patients with advanced cancer and their relatives but is often assessed from only one perspective, namely that of bereaved relatives. This study aimed to gain insight into the quality of care as experienced by patients with advanced cancer and their relatives. Methods A total of 367 patients with stage IV cancer, 242 relatives and 163 bereaved relatives were included from a large prospective, longitudinal study (eQuiPe), which ran from November 2017 until March 2020. Patients and their relatives completed a questionnaire during the last 3 months of the patient’s life. Bereaved relatives completed a questionnaire within six months after the patient’s death. Results At the end of life, patients reported a mean satisfaction with care score of 72/100 (SD 21), and relatives a mean score of 59/100 (SD 28) for the care they received themselves. Continuity with care, the extent to which the care received from different healthcare professionals was coordinated, was associated with higher satisfaction with care in patients (β 2.1, 95% CI 1.6–2.6). Bereaved relatives reported that most patients died peacefully (87%) and at home (73%). Most bereaved relatives (66%) were contacted by a healthcare professional after the patient’s death, but over half were not informed about grief (52%) or the available options for bereavement support (58%), with about 20% reporting they would have appreciated this. Conclusions Quality of end-of-life care was generally perceived as good. This study highlights the importance of good continuity of care as it is associated with higher satisfaction with care in patients. Also, one-fifth of the bereaved relatives reported that they had not been informed about bereavement care despite desiring it, which emphasizes the need for better care for relatives before and after the patient’s death.

NFKB1 genetic variation and allergic rhinitis susceptibility: a study in the Chinese Han population

Protocol of a multi-centre randomized controlled trial to compare pericapsular nerve group block, fascia-iliaca compartment block and femoral nerve block for pain management in patients with a hip fracture in the emergency department (CPFF-ED)

Background Patients presenting in the emergency department (ED) with a hip fracture are often in significant pain. Opioids are frequently prescribed as part of standardized pain protocols, but are associated with many adverse events, especially in the elderly. Regional anaesthesia with a femoral nerve block (FNB), Fascia Iliaca compartment block (FICB) or Pericapsular nerve block (PENG) may provide a reasonable alternative to these patients. We designed a prospective multicentre study to compare the efficacy and safety of these blocks in the ED setting, which has been registered in the Centrale Commissie Mensgebonden Onderzoek (CCMO) Research Portal (NL-OMON57433). Methods and analysis A multicentre Randomized Controlled Trial (RCT), conducted in the ED of 5 hospitals in the Netherlands. A total of 254 adult patients (18+) presenting to the ED with a neck of femur or pertrochanteric femur fracture will be randomised to receive FNB, FICB or PENG in the ED after informed consent has been obtained. Patients will be followed for 12 hours after block placement in the ED. The primary (patient reported) outcome measure is overall patient wellbeing 6–8 hours after block administration (or until operation, whichever comes first) as scored by the validated Quality of Recovery (QoR-15) questionnaire. Secondary outcome measures will be pre-operative pain scores, the cumulative dose of analgesics used during the first 12 hours after block placement or until operation, and complication rates for each of the blocks. Expected results This randomized controlled trial with a superiority design will allow conclusions on which regional nerve block is most effective in improving pre-operative wellbeing in patients with a neck of femur or pertrochanteric fracture. The choice to use a patient reported outcome measure to evaluate block efficacy maximizes clinical relevance. And by allowing the choice of local anaesthetic to the participating hospital (with a 1:1:1 randomization for each participating centre) clinical applicability of our findings is maximized.

Evaluating the RUNOFF01 model’s performance and potential for micro-catchment design

Does generative AI dependence foster creativity? Evidence from sports undergraduates

As an emerging technological tool, generative AI profoundly reshapes higher education teaching and learning processes. Creativity, as a key component of university students’ core competencies, is significantly influenced by technological transformations. However, the mechanisms through which generative AI dependence affects creativity remain underexplored, especially among students majoring in sports. Grounded in Flow Theory, this study investigates the impact of generative AI dependence on the creativity of university students in sports-related disciplines. Using a snowball sampling strategy, the researchers surveyed 453 undergraduates in sports, assessing their self-reported responses across four constructs: generative AI dependence, self-efficacy, flow, and creativity. Partial Least Squares – Structural Equation Modeling was employed to uncover the complex relationships between generative AI dependence and creativity. The results reveal that generative AI dependence has a significant positive direct effect on creativity and an indirect effect mediated by self-efficacy. However, the mediating role of flow was not statistically significant. Furthermore, the multi-group analysis indicates that the positive effects of generative AI dependence on flow and creativity are more pronounced among male students. These findings offer theoretical insights and practical guidance for using generative AI to enhance creativity among students in sports programs.

Whole-genome analysis of CVA4 in children with herpangina in two cities of Shandong, China

Retraction: N-Beats architecture for explainable forecasting of multi-dimensional poultry data

Tissue-specific distribution of cytotoxic triterpene saponins in three Red Sea sea cucumbers

Job competence gap of quantity surveying graduates of higher vocational colleges with a high dimensional measurement error model

The primary aim of this study is to measure the competency gaps between industry expectations and current competencies of recent Chinese quantity surveying graduates from higher vocational colleges, using self-reported data. An innovative measure with a high-dimensional measurement error model was utilized to evaluate competency gaps. We used a quantitative technique with purposive sampling to collect data via a questionnaire survey of quantity surveying graduates with a higher vocational diploma who had worked in China for no more than three years. The findings indicate a competency gap in job matching between quantity surveying higher vocational college graduates’ performance and industry expectations. The personality competency gap is medium, whereas the knowledge, skills, and ethical gaps are minimal. The top three gaps are in business administration, personal image, and innovative insight. The study’s findings highlight the need for training programs focused on management and personality development for quantity surveying higher vocational college graduates. They also emphasize the importance of quantity surveying professionals acquiring interdisciplinary skills through educational programs that combine management with personality attributes such as personal image and innovative insight.

Boosting visible-light photocatalysis with MWCNT-modified TiO2/SiO2/g-C3N4: efficient tetracycline removal in pure and hard water

Abstract In this study, we present a facile synthesis of a TiO 2 /SiO 2 /g-C 3 N 4 /MWCNT composite with dual functionality as both a photocatalyst and adsorbent for the effective elimination of pharmaceutical and dye pollutants from water. The composite was comprehensively characterized using FESEM, EDX, XRD, BET, FTIR, PL, and UV–Vis spectroscopy to elucidate its morphology, crystalline structure, and optical properties. The results revealed a hierarchical flower-like sphere/sheet architecture with an average crystallite size of about 21.3 nm. Multi-walled carbon nanotubes (MWCNTs) acted as visible-light photosensitizers, enhancing the separation efficiency of photogenerated electron–hole pairs. The influence of MWCNT content on photocatalytic activity was systematically investigated. The TiO 2 /SiO 2 /g-C 3 N 4 /MWCNT composite with 11 wt% CNT demonstrated superior visible-light degradation performance, achieving 92% efficiency for a 20 mg/L methylene blue (MB) aqueous solution significantly outperforming the MWCNT-free counterpart. Furthermore, the presence of MWCNTs remarkably improved tetracycline (TC) removal in CaCO 3 -rich water (20 mg/L TC with 300 mg/L CaCO 3 ). The high photocatalytic activity is attributed to the formation of abundant local junctions among g-C 3 N 4 , TiO 2 /SiO 2 , and MWCNTs, which facilitate efficient charge separation and migration through a direct Z-scheme mechanism under visible light illumination.

Hormonal dynamics and transcriptomic regulatory mechanisms during seed dormancy release in Aconitum kusnezoffii

Aconitum kusnezoffii is an important medicinal plant whose seeds exhibit deep physiological dormancy. To elucidate the endogenous hormonal regulation mechanisms and dynamic gene expression patterns during dormancy release, we conducted a comprehensive analysis integrating hormone quantification and transcriptome sequencing on seed samples subjected to cold stratification for 0, 14, and 42 days. Morphological observations indicated that the embryo development rate remained stable during the first 14 days of stratification but increased significantly after 21 days, displaying a sigmoidal growth pattern. This provided a rationale for delineating the dormancy maintenance, release initiation, and pre-germination phases and for selecting key sampling time points. Hormonal profiling revealed a continuous decline in abscisic acid (ABA) content, accompanied by a gradual increase in germination-promoting hormones such as gibberellins (bioactive GA, reported as GA₃-equivalents), indole-3-acetic acid (IAA), brassinolide (BL) and methyl jasmonate (MeJA). The ratios of GA/ABA (GA₃-equivalents), IAA/ABA, and zeatin riboside (ZR; ELISA-based total tZR + cZR equivalents)/ABA increased accordingly, suggesting that dormancy release is orchestrated through a synergistic interplay of multiple hormones. Transcriptome analysis yielded 79,251 Unigenes, with differentially expressed genes significantly enriched in pathways related to energy metabolism, hormone signal transduction, and carbohydrate metabolism. STEM clustering identified five representative temporal expression profiles. Further analysis of hormone-related pathways showed that ABA biosynthesis and signaling components were predominantly active at early stages but declined thereafter, whereas GA and IAA biosynthetic and signaling genes were markedly upregulated during the later stages of dormancy release. These findings highlight the close association between embryo development and hormonal dynamics during seed dormancy release in A. kusnezoffii , and clarify the stage-specific features of hormone synthesis and signaling pathways, providing a theoretical basis for understanding dormancy mechanisms and improving artificial germination techniques.

DQN-empowered energy optimization for wireless powered communication networks

Retraction: Ribosomal L22-like1 (RPL22L1) Promotes Ovarian Cancer Metastasis by Inducing Epithelial-to-Mesenchymal Transition

Efficacy of heat-treated Lacticaseibacillus paracasei PS23 for individuals with long coronavirus disease-19 syndrome: a double-blinded randomized control pilot study

Correction: Effect of photodynamic therapy on choroid of the medial area from optic disc in patients with central serous chorioretinopathy

Physics-guided machine learning framework for RCA concrete by experimental database, modelling, and statistical validation

MicroRNA-488-3p-loaded engineered exosomes inhibit proliferation, migration and invasion of hepatocellular carcinoma by targeting SEC61G

SEC61G is an oncogene in hepatocellular carcinoma (HCC), a common malignant tumor worldwide. MicroRNAs (miRNAs) regulation of oncogenes are available therapeutic strategies being investigated in HCC, but the effective miRNA delivery remains a challenge. Here, we investigated the potential therapeutic effects of miRNA-loaded engineered exosomes in patients with HCC. MiRNAs that could bind to SEC61G were screened using Targetscan, and were verified using HepG2 cells viability after transfecting miRNAs mimic. Five miRNAs binding to SEC61G,among which, miR-651-3p and miR-488-3p mimic significantly inhibited HepG2 cells viability (p < 0.05) and decreased SEC61G protein expression. Then, dual-luciferase reporter assay also confirmed SEC61G as a target of miR-488-3p in HCC. After that, miR-488-3p-loaded engineered exosomes (Exo-miR-488-3p) were isolated from the supernatant of 488-3p-overexpressed cells and identified via nanoparticle tracking analysis, transmission electron microscopy and western blot. In vitro experiments showed that exo-miR-488-3p significantly inhibited proliferation, colony formation, migration and invasion of HepG2 cells than corresponding negative control. In addition, Exo-miR-488-3p tended to induce HepG2 cells apoptosis, though this relationship was not statistically significant. In conclusion, exo-miR-488-3p inhibits the malignant cytological activities in HCC, a possible strategy in the treatment of HCC.

Resolving the contradiction between simulation and experimental results of using gold nanoparticles in proton therapy

Self-reported nursing competence among registered nurses in Jordan: A cross-sectional study

Background Nurses constitute a significant portion of Jordan’s healthcare workforce, and their competence plays a critical role in patient safety and quality of care. Global and national strategies highlight the importance of assessing and improving nursing competence. Despite the critical role of nurses, research on self-reported nursing competence (SRNC) in Jordan remains limited. Aim This study explores self-reported nursing competence (SRNC) among registered nurses (RNs) in Jordan. Methods A cross-sectional study was conducted with a sample of 212 RNs from public, private, and teaching hospitals. Data were collected using the short form of the Nursing Professional Competence (NPC) scale, a 35-item instrument scored on a 7-point Likert scale ranging from 1 (very low competence) to 7 (very high competence), with higher scores indicating greater perceived professional competence. The derived data was analyzed using SPSS software. Results The overall level of self-reported nursing competence was high, with a mean score of 76.69%. Among these six domains, the domain of “documentation and administration of care” scored the highest (78.26%), and “care pedagogy” scored the lowest (75.23%). The items that scored high included “respectful communication,” “clinical follow-up,” and “documentation,” which scored “to a high degree.” The items that scored lower included “psychological needs,” “group-based education,” and “respect for different values,” which scored “to a fairly high degree.”.”. Conclusions The Jordanian nursing community views itself as very competent; however, improvements are needed in particular fields. Continuous learning and training are required in order to retain and upgrade professional competency. The findings highlight the value of mentorship, a supportive work environment, and continuing education. Future research should include peer and supervisor evaluations to validate self-reported competence and better guide workforce development strategies.