With large language models (LLMs) achieving remarkable breakthroughs in natural language processing (NLP) domains, LLM-enhanced recommender systems have received much attention and have been actively explored currently. In this paper, we focus on adapting and empowering a pure large language model for zero-shot and few-shot recommendation tasks. First and foremost, we identify and formulate the lifelong sequential behavior incomprehension problem for LLMs in recommendation domains, i.e., LLMs fail to extract useful information from a textual context of long user behavior sequence, even if the length of context is far from reaching the context limitation of LLMs. To address such an issue and improve the recommendation performance of LLMs, we propose a novel framework, namely Retrieval-enhanced Large Language models (ReLLa) for recommendation tasks in both zero-shot and few-shot settings. For zero-shot recommendation, we perform semantic user behavior retrieval (SUBR) to improve the data quality of testing samples, which greatly reduces the difficulty for LLMs to extract the essential knowledge from user behavior sequences. As for few-shot recommendation, we further design retrieval-enhanced instruction tuning (ReiT) by adopting SUBR as a data augmentation technique for training samples. Specifically, we develop a mixed training dataset consisting of both the original data samples and their retrieval-enhanced counterparts. We conduct extensive experiments on three real-world public datasets to demonstrate the superiority of ReLLa compared with existing baseline models, as well as its capability for lifelong sequential behavior comprehension. To be highlighted, with only less than 10% training samples, few-shot ReLLa can outperform traditional CTR models that are trained on the entire training set (e.g., DCNv2, DIN, SIM).
Background Few real-world studies exist regarding the clinical value of local consolidative therapy (LCT) for oligo-residual disease (ORD) in NSCLC patients treated with immune checkpoint inhibitors. Therefore, we retrospectively evaluated whether LCT could improve the prognosis of NSCL patients with ORD after effective first-line PD-1/PD-L1 inhibitors treatment. Methods A total of 132 patients with metastatic NSCLC who had received first-line PD-1/PD-L1inhibitors-based systemic treatment and developed ORD (defined as residual tumors limited to three organs and five lesions) were included. The LCT group consisted of 41 patients received LCTs for oligo-residual lesions before treatment failure, and the remaining 91 patients, who did not receive local therapies, constituted the non-LCT group. The progression-free survival (PFS) and overall survival (OS) of the two groups were analyzed. Results With a median follow-up of 12.0 months, 86 patients developed progressive disease and 42 patients died. Compared with the non-LCT group, LCT group exhibited significant longer progression-free survival (PFS) (median 11.0 vs. 7.0 months, P=0.017) and overall survival (OS) (median 26.0 vs. 17.0 months, P=0.003). Multivariable analysis demonstrated that LCT was an independent predictor of prolonged PFS (HR=0.606, 95% CI=0.370–0.964, P=0.035) and OS (HR=0.467, 95% CI=0.229–0.949, P=0.035). Subgroup analysis revealed that the dominant population considerably benefited from LCT in terms of PFS and OS included patients with 1-2 residual tumor sites (mPFS: 11.0 vs. 7.0 months, P=0.013; mOS: 23.0 vs. 17.0 months, P=0.018) and those with high PD-L1 expression (mPFS: 13.0 vs. 7.0 months, P=0.018; mOS: 34.0 vs. 16.0 months, P=0.030). In addition, the All-LCT group had significantly longer PFS (mPFS 16.0 vs. 7.0, P=0.002) and OS (mOS 28.0 vs. 17.0, P= 0.002) than did the non-LCT group. However, patients who received LCT to only some of their lesions had not experienced improvements in PFS (P=0.546) or OS (P=0.198). Conclusion LCT may provide extra survival benefits among patients with oligo-residual NSCLC after effective first-line PD-1/PD-L1 inhibitors treatment, particularly in those patients with one or two residual lesions, high PD-L1 expression, or who had received LCT for all lesions. LCT may be a novel treatment option for this specific population.
Oncogene StarD4 had the function of promoting proliferation and metastasis of triple-negative breast cancer (TNBC), but its clinical value and molecular mechanism are unknown. This paper found that StarD4 was highly expressed in cancer tissues of TNBC patients, and higher expression level of StarD4 in TNBC patient resulted in poorer prognosis. Based on transcriptomics of MDA-MB-231 cell model, the results of bioinformatics analysis showed that down-regulated expression level of StarD4 led to overall downregulation of cholesterol-relative genes and significant enrichment of cancer mechanism and pathway. Further analysis and investigation verified that StarD4 might cross-promote the protein stability of receptor ITGA5 through the cholesterol pathway to enhance TNBC progression, which provides guidance for clinical application of TNBC diagnosis and treatment.StarD4具有促三阴乳腺癌（TNBC）增殖转移功能，但临床价值和分子机制未明。本文发现StarD4在TNBC癌组织高表达，且高表达患者生存预后不良。通过TNBC细胞模型的转录组检测和分析发现：StarD4敲减表达引发胆固醇通路基因的整体下调和TNBC肿瘤通路的显著富集。进一步分析验证了StarD4可能通过胆固醇通路交叉调控细胞膜受体ITGA5的蛋白稳定性，发挥促癌的分子机制，为临床TNBC的诊疗应用提供了指导。.
This study aimed to investigate the serial-multiple mediation effect of professional identity, psychological capital (PsyCap), work-related stress, and work-related wellbeing among intensive care unit (ICU) nurses in China. The cross-sectional survey was conducted from January 2017 to May 2017 in two Grade III A general hospitals (with more than 2000 beds) in Jining, Shandong Province, China. Cluster sampling was used to recruit participants from the two hospitals. A total of 330 ICU nurses participated in the study. The nurses’ work stress scale, Chinese nurse’s professional identity scale, the PsyCap questionnaire, and Chinese work-related wellbeing scale were used to collect the data. Descriptive analysis, independent-samples t- test, one-way analysis of variance, Pearson correlation analysis, linear regression analysis, and structural equation modeling were used to analyze the data ( P < 0.05 was considered statistically significant). The average score for the work-related wellbeing of ICU nurses was 85.91 ± 13.94. Work-related stress, professional identity, and PsyCap correlated significantly with work-related wellbeing. The major predictors of work-related wellbeing were PsyCap, work-related stress, professional identity, and monthly salary. The serial-multiple mediation effects of professional identity and PsyCap in the relationship between work-related stress and work-related wellbeing were statistically significant. Positive professional identity and PsyCap were sequentially associated with decreased work-related stress, which in turn was related to increased work-related wellbeing among ICU nurses. Therefore, this study aims to explore the impact of ICU nurses’ work-related stress on work-related wellbeing, as well as the mediating effect of professional identity and PsyCap. It is hoped that hospital care managers will pay attention to the mental health of ICU nurses, increase their professional identity, and reduce work-related stress to improve the quality of the ICU nursing service and stabilize nursing work.
Our study was to investigate the effects of interleukin-6 (IL-6) polymorphisms (rs2069837 and rs17147230) on the risk for hepatocellular carcinoma (HCC).A total of 226 HCC cases and 220 healthy controls were admitted into the study and genomic DNA was extracted from the peripheral blood. The genotyping was conducted by the method of polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). Odds ratio (OR) and 95% confidence interval (CI) were used to evaluate the relationship of IL-6 rs2069837 and rs17147230 polymorphisms with HCC susceptibility.The frequency of GG genotype of rs2069837 was higher in HCC patients, compared with controls (P < 0.05). Moreover, the results indicated that GG genotype was related with increased risk for HCC (OR = 2.303, 95% CI = 1.056-5.025). Similarly, the risk for G allele carriers was higher than that of A allele (OR = 1.392, 95% CI = 1.046-1.852). For rs17147230, TT genotype showed strong effect on HCC susceptibility (OR = 2.089, 95% CI = 1.135-3.845) and T allele appeared to be a risk factor for HCC (OR = 1.326, 95% CI = 1.010-1.740). Further analysis showed that G-T haplotype was associated with increased risk for HCC (OR = 3.125, 95% CI = 1.845-5.294, P = 0.000).IL-6 rs2069837 as well as rs17147230 were associated with HCC susceptibility. In addition, G-T haplotype also served as a genetic-susceptibility factor for HCC.
Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI
This study focuses on the splitter blade pump–turbine as the research object to analyze the problems of hump characteristics and the hysteresis effect. We simulated the operation of the pump condition with small opening of the guide vane, analyzed the hydraulic loss by using the entropy production theory and entropy wall function, and investigated the study of internal flow transfer characteristics. In this paper, it was first verified that the maximum error of the energy loss calculated by the pressure method and the entropy production method was less than 6% for the working zone. From the quantified energy loss results, a significant instability feature was observed in the 0.65 QBEP–0.9 QBEP operating interval, accompanied by the phenomenon of the non-overlapping of the characteristic curves. The results show that the hump characteristic with hysteresis effect also exists in the splitter blade pump–turbine. The percentage of energy loss in the hump zone is in descending order of runner, guide vanes, spiral casing, and draft tube, but this changes again at low flow rates. By analyzing the high-entropy production region, it was found that the high-hydraulic-loss region is mainly distributed at the trailing edge of the long blade in the vane-less space, which is different from the traditional runner.
Large language models (LLMs) have achieved remarkable progress in the field of natural language processing (NLP), demonstrating remarkable abilities in producing text that resembles human language for various tasks. This opens up new opportunities for employing them in recommender systems (RSs). In this paper, we specifically examine the sample efficiency of LLM-enhanced recommender systems, which pertains to the model's capacity to attain superior performance with a limited quantity of training data. Conventional recommendation models (CRMs) often need a large amount of training data because of the sparsity of features and interactions. Hence, we propose and verify our core viewpoint: Large Language Models Make Sample-Efficient Recommender Systems. We propose a simple yet effective framework (i.e., Laser) to validate the viewpoint from two aspects: (1) LLMs themselves are sample-efficient recommenders; and (2) LLMs, as feature generators and encoders, make CRMs more sample-efficient. Extensive experiments on two public datasets show that Laser requires only a small fraction of training samples to match or even surpass CRMs that are trained on the entire training set, demonstrating superior sample efficiency.
Breast cancer (BRCA) is one of the most common malignancies encountered in women worldwide. Lipid metabolism has been found to be involved in cancer progression. Steroidogenic acute regulatory protein‑related lipid transfer 4 (STARD4) is an important cholesterol transporter involved in the regulatory mechanism of intracellular cholesterol homeostasis. However, to the best of our knowledge, the molecular functions of STARD4 in BRCA are unclear. Immunohistochemical staining and public dataset analysis were performed to investigate the expression levels of STARD4 in BRCA. In the present study, high expression of STARD4 was identified in BRCA samples and higher STARD4 expression was significantly associated with shorter distant metastasis‑free survival time in patients with BRCA, which indicated that STARD4 may be associated with BRCA progression. Cell cytometry system Celigo® analysis, Cell Counting K‑8 assays, flow cytometry, wound healing assays and transwell assays were used to investigate the effects of STARD4 knockdown on proliferation, cell cycle, apoptosis and migration in BRCA cells. Loss‑of‑function assays demonstrated that STARD4 acted as an oncogene to promote proliferation and cell cycle progression, while suppressing apoptosis in BRCA cells in vitro and in vivo. Furthermore, knockdown of STARD4 significantly suppressed BRCA metastasis. To assess the mechanism of action of STARD4, microarray analysis was performed following STARD4 knockdown in MDA‑MB‑231 cells. The data were analyzed in detail using bioinformatics, and a series of genes, including E74 like ETS transcription factor 1, cAMP responsive element binding protein 1 and p21 (RAC1) activated kinase 2, which have been previously reported to be crucial genes implicated in the malignant phenotype of cancer cells, were identified to be regulated by STARD4. Loss‑of function assays demonstrated that knockdown of STARD4 suppressed BRCA proliferation and migration. These findings suggested that STARD4 had an oncogenic effect in human BRCA progression.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)