Objective To develop a system for retrieving information from clinical records of Traditional Chinese Medicine(TCM)based on generative large language models(LLMs).Methods Applicational needs of the system were analyzed,and entity types to be retrieved were identified.The functions,workflows,and architecture of the system were designed by combining the automatic retrieval capabilities of LLMs with human-in-the-loop(HITL).The software was developed using such frameworks as vLLM and Node.js.Interaction of multiple commercial/open source LLMs was implemented using OpenAI-compatible interfaces.The quality of information retrieved from LLMs was enhanced by prompt engineering.Results This system supported task allocation,automatic retrieval of structured information,and manual review.To evaluate its performance,the moonshot-v1-8k model was used to retrieve clinical records of TCM before manual edition was performed.Combining large language model pre-annotation with meticulous annotator edits improved accuracy by 26.6％compared to the BERT-BiLSTM-CRF model,and enhanced extraction efficiency by 1.6-fold relative to purely manual methods.Conclusion General generative LLMs can retrieve a wide range of entity information from TCM records with high accuracy and scalability.The design and implementation of this system approach may provide a useful reference for developing other biomedical information retrieval systems.

Objective: To predict and screen potential biomarkers of systemic lupus eythematosus(SLE) based on machine learning algorithms and structural biology, and to reveal their mechanisms of action and to provide new targets for disease diagnosis and treatment. Methods: Four machine learning algorithms, random forest(RF), eXtreme gradient boosting(XGBoost), support vector machine(SVM), least absolute shrinkage and selection operator(LASSO), were used to analyze the gene expression data of SLE patients in GEO(datasets: GSE121239 and GSE11907) to analyze the gene expression data of SLE patients and screen key markers. Peripheral blood single nucleated cells(PBMCs) from SLE patients were collected and RT-qPCR was used to detect differential gene expression levels. Subsequently, GSEA enrichment analysis was used to identify biomarker-related pathways. CIBERSORT immune infiltration analysis and protein interactions network were applied to calculate the sample immune cell infiltration abundance. Single-cell data were analyzed for gene expression specificity in immune cells. Interaction relationships in combination with AlphaFold3(AF3) were predicted. Results: Multiple algorithms were screened together to identify the unique marker gene HERC5 , and expression analysis of multiple datasets showed that HERC5 was highly expressed in SLE compared to the normal group (P < 0. 05) , and RT⁃qPCR verified the same trend (P = 0. 006 2) . Functional enrichment analysis identified the major pathway promoted by HERC5 in SLE as the interferon receptor signalling pathway (P < 0. 05) . Immune infiltration analysis showed that HERC5 was closely associated with immune cells (Neutrophils : r = 0. 39 , P < 0. 05 ; Memory B cells : r = 0. 33 , P < 0. 05 ; Activated dendritic cell : r = 0. 52 , P < 0. 05) . Most HERC5 ⁃related interacting proteins were associated with SLE ,and potential transcription factors of HERC5 and its related genes were also significantly associated with immune responses. Conclusion The HERC5 gene is an important biomarker for SLE , which upregulates the interferon pathway to promote SLE progression and provides a new target for SLE diagnosis and treatment.

Objective To analyze the hotspots and developments in the field of language model-assisted artificial intelli-gence(Al)for antibody design and optimization in order to provide reference for research on development of antibodies.Methods By using CiteSpace software,hotspots of research were analyzed based on literature retrieved from the Web of Science,PubMed,and Scopus databases,focusing on three pivotal areas of research related to antibody design and optimization:the construction of pre-trained language models for antibodies,the generation of antibody sequences,and the prediction of three-dimensional structures of antibodies.In addition,this analysis reviewed the major advances in each of the specified research tasks,focusing on the delineation of similarities and differences across studies and dominating challenges in this field.Results From 2019(10 publications)to 2023(89 publications),the scale of and interest in this field kept increasing.Hotspots involved leveraging language models to assist the design or optimization of humanized,high-affinity,and highly specific antibodies.Within each research,methods were characterized by the diversity of model architectures,consistency of training data,and variations in training strategies.Challenges to the field included sparse antigen data,computational power limitations,and insufficient integration of wet and dry lab experiments.Conclusion Research in language model-assisted Al antibody design and optimization is gaining momentum and proves fruitful.However,researchers should be alert to the inadequate attention to antigen-antibody interactions and insufficient integration of experimental and computational validation,conduct more in-depth research and expand applications.

Background: Both type 1 diabetes mellitus (T1DM) and metabolic syndrome (MetS) are associated with an elevated risk of morbidity and mortality yet with increasing heterogeneity. This study primarily aimed to evaluate the prevalence of MetS among adult patients with T1DM in China and investigate its associated risk factors, and relationship with microvascular complications. Methods: We included adult patients who had been enrolled in the Guangdong T1DM Translational Medicine Study conducted from June 2010 to June 2015. MetS was defined according to the updated National Cholesterol Education Program criterion. Logistic regression models were used to estimate the odds ratio (OR) for the association between MetS and the risk of diabetic kidney disease (DKD) and diabetic retinopathy (DR). Results: Among the 569 eligible patients enrolled, the prevalence of MetS was 15.1%. While female gender, longer diabetes duration, higher body mass index, and glycosylated hemoglobin A1c (HbA1c) were risk factors associated with MetS (OR, 2.86, 1.04, 1.14, and 1.23, respectively), received nutrition therapy education was a protective factor (OR, 0.46). After adjustment for gender, age, diabetes duration, HbA1c, socioeconomic and lifestyle variables, MetS status was associated with an increased risk of DKD and DR (OR, 2.14 and 3.72, respectively; both P<0.05). Conclusion Although the prevalence of MetS in adult patients with T1DM in China was relatively low, patients with MetS were more likely to have DKD and DR. A comprehensive management including lifestyle modification might reduce their risk of microvascular complications in adults with T1DM.

ood long-term efficacy for cirrhosis patients with portal hypertension was a useful treatment for these patients.

This study investigated the inhibitory effect of the extract of fungi of Huaier (EFH) on the growth of hepatocellular carcinoma (HCC) cells.Hep-G2 cells,a human HCC cell line,were cultured in DMEM containing 10% fetal bovine serum and treated with EFH of different concentrations (1,2,4,8 mg/mL) for 24,48 and 72 h respectively.The apoptosis rate of the cells was flow cytometrically measured.Thirty-six tumor-bearing New Zealand rabbits were randomly divided into 3 groups:group A (control group),in which the rabbits were infused with 0.2 mL/kg normal saline via the hepatic artery;group B (transhepatic artery chemoembolization [TACE] group),in which the rabbits were given lipiodol at 0.2 mL/kg plus MMC at 0.5 mg/kg via the hepatic artery;group C (TACE +EFH group),in which EFH (500 mg/kg) were orally administered after TACE.Two weeks after TACE,the rabbits were sacrificed and the implanted tumors were sampled.The tumor volume and the necrosis rate were determined.The tumor tissues were immunohistochemically detected for the expressions of factor Ⅷ,VEGF,P53,Bax and Bcl-2.The microvessel density (MVD) was calculated by counting the factor Ⅷ-positive endothelial cells.Our results showed that after treatment with EFH,the apoptosis rate of Hep-G2 cells was enhanced in a concentration- and time-dependent manner.Two weeks after the treatment,the average tumor volume,the necrosis rate and the growth rate of the implanted tumor in group C were significantly different from those in groups A and B (P<0.05).MVD and VEGF expressions were significantly decreased in the group C when compared with those in groups B (P<0.05 for all).The Bax expression was weakest in group A and strongest in group C.The expressions of P53 and Bcl-2 were minimal in group C and maximal in group A.There were significant differences in the expressions of P53,Bax and Bcl-2 among the 3 groups (P<0.05 for all) and there was significant difference between group B and group C (P<0.05).It was concluded that EFH could suppress not only the growth of HCC cells but also tumor angiogenesis and it can induce the apoptosis of HCC cells.EFH serves as an alternative for the treatment of HCC.