Purpose/Significance To construct a named entity corpus of traditional Chinese medicine(TCM)ancient records,and to improve the recognition accuracy and applicability of the general domain named entity recognition(NER)model in the field of TCM ancient records.Method/Process Annotation standards for entities in TCM ancient records are formulated,and 2 384 Xin'an medical records are annotated.A RoBERTa-BiLSTM-CRF model is developed,and word vectors with semantic features are generated using the RoBERTa pre-trained language model.The BiLSTM-CRF model is used to learn the global semantic features of sequences and decode and output the optimal label sequence.Dictionary and rule features are incorporated to enhance the model's capability to recognize entity boundaries and categories.Result/Conclusion The model shows a good recognition effect on the named entity corpus of Xin'an medical cases.Integration of domain terminology dictionaries and rule-based features improves the overall Fl score to 72.8％.

OBJECTIVE To develop an artificial intelligence-based intelligent auxiliary diagnosis and treatment system for Xin'an medicine to address the challenges of integrating ancient Xin'an medical case records into modern clinical applications.METHODS The project involved structuring and standardizing case records from ancient texts of Xin'an medicine to build a compre-hensive Xin'an medicine database.Advanced techniques,such as data annotation,entity relationship extraction,and data mining,were applied to create a Xin'an medicine knowledge base.Furthermore,a knowledge graph of Xin'an medicine was constructed using techniques for knowledge acquisition,integration,storage,and graph-based question-answering,improving the efficiency of knowl-edge organization and retrieval.The LangChain framework was utilized to connect the Xin'an medicine knowledge base to a large lan-guage model,enabling a model-driven local knowledge base question-answering system.RESULTS The study successfully estab-lished a systematic and standardized knowledge base for Xin'an medical case records.The application of knowledge graph technology provided a clear visualization of Xin'an medicine's knowledge structure,and the development of an intelligent question-answering module significantly improved the efficiency of knowledge management and retrieval.The local knowledge base question-answering sys-tem,powered by a large language model and based on Xin'an medicine's theoretical and practical expertise,delivered accurate diag-nostic and treatment support,promoting the heritage and innovation of Xin'an medicine.CONCLUSION This research validates the feasibility of modernizing traditional medical texts and provides an innovative approach to knowledge development and clinical applica-tion in Chinese medicine.The findings have significant academic value and promising clinical implications.

OBJECTIVE To develop an artificial intelligence-based intelligent auxiliary diagnosis and treatment system for Xin'an medicine to address the challenges of integrating ancient Xin'an medical case records into modern clinical applications.METHODS The project involved structuring and standardizing case records from ancient texts of Xin'an medicine to build a compre-hensive Xin'an medicine database.Advanced techniques,such as data annotation,entity relationship extraction,and data mining,were applied to create a Xin'an medicine knowledge base.Furthermore,a knowledge graph of Xin'an medicine was constructed using techniques for knowledge acquisition,integration,storage,and graph-based question-answering,improving the efficiency of knowl-edge organization and retrieval.The LangChain framework was utilized to connect the Xin'an medicine knowledge base to a large lan-guage model,enabling a model-driven local knowledge base question-answering system.RESULTS The study successfully estab-lished a systematic and standardized knowledge base for Xin'an medical case records.The application of knowledge graph technology provided a clear visualization of Xin'an medicine's knowledge structure,and the development of an intelligent question-answering module significantly improved the efficiency of knowledge management and retrieval.The local knowledge base question-answering sys-tem,powered by a large language model and based on Xin'an medicine's theoretical and practical expertise,delivered accurate diag-nostic and treatment support,promoting the heritage and innovation of Xin'an medicine.CONCLUSION This research validates the feasibility of modernizing traditional medical texts and provides an innovative approach to knowledge development and clinical applica-tion in Chinese medicine.The findings have significant academic value and promising clinical implications.

Objective To investigate the risk factors of electrocoagulation syndrome after endoscopic submucosal dissection (ESD) in patients with colorectal lesions.Methods Clinical data of 145 patients with colorectal mucosal lesions undergoing ESD in People's Hospital of Wuhan University between September 2010 and September 2015 were retrospectively studied.Results Among 45 patients,post endoscopic submucosal dissection electrocoagulation syndrome (PEECS) was developed in 32 cases (22％).The median age in PEECS group was higher (t =-5.783,P =0.000),the median lesion size was larger(t =-5.590,P =0.000),the median length of hospital stay was longer (t =-6.841,P =0.000) than those in non-PEECS group.Univariate regression analysis showed PEECS was associated with the age,lesion size,lesion location,length of hospital stay,malignant tumor,polyps type,resection modality.Multivariable logistic regression analysis showed that the independent risk factors for the development of electrocoagulation syndrome were age ＞65 year (OR =1.123,95％ CI:1.013-1.244,P =0.027),lesion size ＞ 3.5 cm (OR =1.173,95％ CI:1.015-1.357,P =0.031),malignant tumor (OR =3.498,95 ％ CI:1.460-8.379,P =0.005),hospital stay ＞ 10 d (OR =2.480,95％ CI:1.346-4.569,P =0.004),non-rectal lesions (OR =12.612,95％ CI:3.446-46.157,P =0.000).Conclusion Attention should be paid for colorectal lesion patients with high risk of PEECS,when endoscopic submucosal dissection is performed.

Objective To explore the association between patatin-like phospholipase domaincontaining protein 3 (PNPLA3) gene rs738409 polymorphism and the susceptibility of non-alcoholic fatty liver disease (NAFLD).Methods Data bases were comprehensively searched to retrace all the related studies on the association between PNPLA3 gene rs738409 polymorphism and susceptibility.Of NAFLD,the pooled OR with 95％ CI of the association between PNPLA3 gene rs738409 polymorphism and NAFLD susceptibility were performed using different genetic models.Subgroup analysis based on the source of population and sensitivity analysis was performed to detect the stability of results.Results 28 original studies with 6 216 patients and 8 218 controls were involved in the final combination of data.Findings from the meta-analyses showed that there were strong associations between PNPLA3 gene rs738409 polymorphism and the susceptibility of NAFLD,under different genetic model comparisons [GG vs.CC:OR=2.42,95％CI:1.83-3.21,P＜0.001 ;CG vs.CC:OR=1.28,95％CI:1.15-1.43,P＜0.001 ; CG+GG vs.CC:OR=1.31,95％CI:1.17-1.46,P＜ 0.001 ; GG vs.CC + GC:OR=2.26,95％CI:1.76-2.90,P＜0.001].Similar results were found in both Asian and Caucasian populations.Conclusion Results from the Meta-analysis strongly suggested that there appeared significant association between PNPLA3 gene rs738409 polymorphism and the susceptibility of NAFLD.

Objective To study whether matrix metalloproteinases-9 (MMP)-1562C/T (rs3918242) and MMP-2-1306C/T (rs243865) were associated with the susceptibility on nonalcoholic fatty liver disease (NAFLD) and the interactions between the two factors and central obesity.Methods Genotypes of 545 patients and 636 subjects with NAFLD under control were examined by polymerase chain reaction-based restriction fragment length polymorphism (PCR-RFLP).Unconditional logistic regression (ULR) was performed to assess the NAFLD risk.The geneenvironment interactions on the risk of NAFLD were explored by generalized multifactor dimensionality reduction (GMDR) and ULR methods.Results Results from the case-control analysis indicated that there was an increased risk of developing NAFLD for MMP-9 rs3918242 (TT/CT) genotype carriers,when compared with the non-carriers (CC),with OR=1.67,95％ CI:1.32-2.12,P=0.001;Adjusted OR=1.65,95％CI:1.31-2.01 (P=0.008).However,risk reduction of NAFLD was found when MMP-2 rs243865 (TT/CT) genotype carriers compared with the non-carriers (CC),with OR=0.68,95％CI:0.53-0.86,P=0.001;with adjusted OR=0.66,95％CI:0.49-0.90 (P=0.007).Data from the GMDR showed that gene-environment interaction among rs3918242 and central obesity on the risk of NAFLD might be significant (P=0.001).By using the ULR method,subjects as central obesity-positive but with genotype CT/TT,appeared having 4.50 (95％ CI:2.78-7.17,P =0.007) times risk of NAFLD,when compared to the central obesity-negative subjects with genotype CC after adjusting for the covariates.Conclusion MMP-9 rs3918242,MMP-2 rs243865 were associated with risk of NAFLD while both rs3918242 and central obesity showing synergistic effects on the risk of the NAFLD.