GastroBot: a Chinese gastrointestinal disease chatbot based on the retrieval-augmented generation.

Introduction: Large Language Models (LLMs) play a crucial role in clinical information processing, showcasing robust generalization across diverse language tasks. However, existing LLMs, despite their significance, lack optimization for clinical applications, presenting challenges in terms of illusions and interpretability. The Retrieval-Augmented Generation (RAG) model addresses these issues by providing sources for answer generation, thereby reducing errors. This study explores the application of RAG technology in clinical gastroenterology to enhance knowledge generation on gastrointestinal diseases. Methods: We fine-tuned the embedding model using a corpus consisting of 25 guidelines on gastrointestinal diseases. The fine-tuned model exhibited an 18% improvement in hit rate compared to its base model, gte-base-zh. Moreover, it outperformed OpenAI's Embedding model by 20%. Employing the RAG framework with the llama-index, we developed a Chinese gastroenterology chatbot named "GastroBot," which significantly improves answer accuracy and contextual relevance, minimizing errors and the risk of disseminating misleading information. Results: When evaluating GastroBot using the RAGAS framework, we observed a context recall rate of 95%. The faithfulness to the source, stands at 93.73%. The relevance of answers exhibits a strong correlation, reaching 92.28%. These findings highlight the effectiveness of GastroBot in providing accurate and contextually relevant information about gastrointestinal diseases. During manual assessment of GastroBot, in comparison with other models, our GastroBot model delivers a substantial amount of valuable knowledge while ensuring the completeness and consistency of the results. Discussion: Research findings suggest that incorporating the RAG method into clinical gastroenterology can enhance the accuracy and reliability of large language models. Serving as a practical implementation of this method, GastroBot has demonstrated significant enhancements in contextual comprehension and response quality. Continued exploration and refinement of the model are poised to drive forward clinical information processing and decision support in the gastroenterology field.

ChatGPT Combining Machine Learning for the Prediction of Nanozyme Catalytic Types and Activities.

The design of nanozymes with superior catalytic activities is a prerequisite for broadening their biomedical applications. Previous studies have exerted significant effort in theoretical calculation and experimental trials for enhancing the catalytic activity of nanozyme. Machine learning (ML) provides a forward-looking aid in predicting nanozyme catalytic activity. However, this requires a significant amount of human effort for data collection. In addition, the prediction accuracy urgently needs to be improved. Herein, we demonstrate that ChatGPT can collaborate with humans to efficiently collect data. We establish four qualitative models (random forest (RF), decision tree (DT), adaboost random forest (adaboost-RF), and adaboost decision tree (adaboost-DT)) for predicting nanozyme catalytic types, such as peroxidase, oxidase, catalase, superoxide dismutase, and glutathione peroxidase. Furthermore, we use five quantitative models (random forest (RF), decision tree (DT), Support Vector Regression (SVR), gradient boosting regression (GBR), and fully connected deep neuron network (DNN)) to predict nanozyme catalytic activities. We find that GBR model demonstrates superior prediction performance for nanozyme catalytic activities (R2 = 0.6476 for Km and R2 = 0.95 for Kcat). Moreover, an open-access web resource, AI-ZYMES, with a ChatGPT-based nanozyme copilot is developed for predicting nanozyme catalytic types and activities and guiding the synthesis of nanozyme. The accuracy of the nanozyme copilot's responses reaches more than 90% through the retrieval augmented generation. This study provides a new potential application for ChatGPT in the field of nanozymes.

Bibliometric and visual analysis of esophagogastric junction cancer research from 2002 to 2021.

Numerous studies related to esophagogastric junction cancer (EGC) have been published, and bibliometric analysis of these publications may be able to identify research hotspots and frontiers of EGC. Studies published on EGC between 2002 and 2021 were retrieved from the Web of Science Core Collection. The collaboration network of countries/regions, institutions, authors, co-citation network of journals, co-occurrence network, and overlay visualization of keywords were analyzed using the VOSviewer software. Cluster and timeline analyses of references were performed using the CiteSpace software. A total of 5109 English articles were published across 691 journals by authors affiliated with 4727 institutions from 81 countries/regions. The annual number of publications related to EGC research has exhibited an increasing trend. The United States, China, and Japan emerged as the top 3 prolific countries/regions. Institutions in the United States, Japan, and South Korea exhibited significant collaboration with one another. Diseases of the Esophagus was the most prolific journal, and Annals of Surgical Oncology, World Journal of Gastroenterology, and Gastric Cancer had also published more than 100 studies. Jaffer A Ajani was the most productive author while David Cunningham ranked the first in terms of total citations and average citations per article. Barrett's esophagus, gastroesophageal reflux disease, Helicobacter pylori, and obesity were common topics in earlier research, and recent years had seen a shift towards the topics of immunotherapy, targeted therapy, and neoadjuvant chemotherapy. In conclusion, growing attention is paid to EGC research, especially in terms of immunotherapy, targeted therapy, and neoadjuvant chemotherapy.

MSLN induced EMT, cancer stem cell traits and chemotherapy resistance of pancreatic cancer cells.

Chemoresistance is one of the main reasons for poor prognosis of pancreatic cancer. The effects of mesothelin (MSLN) on chemoresistance in pancreatic cancer are still unclear. We aim to investigate potential roles of MSLN in chemoresistance and its relationship with proliferation, epithelial-mesenchymal transition (EMT) and cancer stemness of pancreatic cancer cells. Human pancreatic cancer cell lines ASPC-1 and Mia PaCa-2 with high and low expression of MSLN, respectively, were selected. The ASPC-1 with MSLN knockout (KO) and Mia PaCa-2 of MSLN overexpression (OE) were generated. The effects of MSLN on cell phenotypes, expression of EMT-related markers, clone formation, tumor sphere formation, and pathologic role of MSLN in tumorigenesis were detected. Sensitivity of tumor cells to gemcitabine was evaluated. The results showed that adhesion, proliferation, migration and invasion were decreased significantly in ASPC-1 with MSLN KO, whereas increased significantly in Mia PaCa-2 with MSLN OE. The size and the number of clones and tumor spheres were decreased in ASPC-1 with MSLN KO, and increased in Mia PaCa-2 with MSLN OE. In xenograft model, tumor volume was decreased (tumor grew slower) in MSLN KO group compared to control group, while increased in MSLN OE group. Mia PaCa-2 with MSLN OE had a higher IC50 of gemcitabine, while ASPC-1 with MSLN KO had a lower IC50. We concluded that MSLN could induce chemoresistance by enhancing migration, invasion, EMT and cancer stem cell traits of pancreatic cancer cells. Targeting MSLN could represent a promising therapeutic strategy for reversing EMT and chemoresistance in pancreatic cancer cells.

CPMI-ChatGLM: parameter-efficient fine-tuning ChatGLM with Chinese patent medicine instructions.

Chinese patent medicine (CPM) is a typical type of traditional Chinese medicine (TCM) preparation that uses Chinese herbs as raw materials and is an important means of treating diseases in TCM. Chinese patent medicine instructions (CPMI) serve as a guide for patients to use drugs safely and effectively. In this study, we apply a pre-trained language model to the domain of CPM. We have meticulously assembled, processed, and released the first CPMI dataset and fine-tuned the ChatGLM-6B base model, resulting in the development of CPMI-ChatGLM. We employed consumer-grade graphics cards for parameter-efficient fine-tuning and investigated the impact of LoRA and P-Tuning v2, as well as different data scales and instruction data settings on model performance. We evaluated CPMI-ChatGLM using BLEU, ROUGE, and BARTScore metrics. Our model achieved scores of 0.7641, 0.8188, 0.7738, 0.8107, and - 2.4786 on the BLEU-4, ROUGE-1, ROUGE-2, ROUGE-L and BARTScore metrics, respectively. In comparison experiments and human evaluation with four large language models of similar parameter scales, CPMI-ChatGLM demonstrated state-of-the-art performance. CPMI-ChatGLM demonstrates commendable proficiency in CPM recommendations, making it a promising tool for auxiliary diagnosis and treatment. Furthermore, the various attributes in the CPMI dataset can be used for data mining and analysis, providing practical application value and research significance.

Clinical study on the necessity and feasibility of routine MRCP in patients with cholecystolithiasis before LC.

BACKGROUND: In the past quite a long time, intraoperative cholangiography(IOC)was necessary during laparoscopic cholecystectomy (LC). Now magnetic resonance cholangiopancreatography (MRCP) is the main method for diagnosing common bile duct stones (CBDS). Whether MRCP can replace IOC as routine examination before LC is still inconclusive. The aim of this study was to analyze the clinical data of patients undergoing LC for cholecystolithiasis, and to explore the necessity and feasibility of preoperative routine MRCP in patients with cholecystolithiasis. METHODS: According to whether MRCP was performed before operation, 184 patients undergoing LC for cholecystolithiasis in the Department of General Surgery, Beijing Shijitan Hospital, Capital Medical University from January 1, 2017 to December 31, 2018 were divided into non-MRCP group and MRCP group for this retrospective study. The results of preoperative laboratory test, abdominal ultrasound and MRCP, biliary related comorbidities, surgical complications, hospital stay and hospitalization expenses were compared between the two groups. RESULTS: Among the 184 patients, there were 83 patients in non-MRCP group and 101 patients in MRCP group. In MRCP group, the detection rates of cholecystolithiasis combined with CBDS and common bile duct dilatation by MRCP were higher than those by abdominal ultrasound (P < 0.05). The incidence of postoperative complications in non-MRCP group (8.43%) was significantly higher (P < 0.05) than that in MRCP group (0%). There was no significant difference in hospital stay (P > 0.05), but there was significant difference in hospitalization expenses (P < 0.05) between the two groups. According to the stratification of gallbladder stone patients with CBDS, hospital stay and hospitalization expenses were compared, and there was no significant difference between the two groups (P > 0.05). CONCLUSIONS: The preoperative MRCP can detect CBDS, cystic duct stones and anatomical variants of biliary tract that cannot be diagnosed by abdominal ultrasound, which is helpful to plan the surgical methods and reduce the surgical complications. From the perspective of health economics, routine MRCP in patients with cholecystolithiasis before LC does not increase hospitalization costs, and is necessary and feasible.

3-sulfonyloxyaryl(mesityl)iodonium triflates as 1,2-benzdiyne precursors with activation via ortho-deprotonative elimination strategy.

Benzyne has long captivated the attention of chemists and has gained numerous synthetic achievements. Among typical benzyne generation methods, removal of two vicinal substituents from 1,2-difunctionalized benzenes, i.e., Kobayashi's protocol, are prevailing, while ortho-deprotonative elimination from mono-substituted benzene lags far behind. Despite the advantages of atom economy and ready achievability of precursors, a bottle neck for ortho-deprotonative elimination strategy resides in the weak acidity of the ortho-hydrogen, which normally demands strong bases as the activating reagents. Here, an efficient aryne generation protocol is developed, where ortho-deprotonative elimination on 3-sulfonyloxyaryl(mesityl)iodonium triflates occurs under mild conditions and the generated 3-sulfonyloxyarynes can serve as efficient 1,2-benzdiyne synthons. This array of 1,2-benzdiyne precursors can be conveniently prepared with high functional group tolerance, and densely substituted scaffolds can be accessed as well. Carbonate and fluoride salts are found to serve as efficient activating reagents, which are the weakest bases used in ortho-deprotonative elimination strategies. Particularly, this scaffold has predictable chemoselective generation of the designated aryne intermediates. The success of this ortho-deprotonative elimination protocol sets up a unique platform with a broad spectrum of synthetic applications.

A Deep Neural Network for Gastric Cancer Prognosis Prediction Based on Biological Information Pathways.

Background: Gastric cancer (GC) is one of the deadliest cancers in the world, with a 5-year overall survival rate of lower than 20% for patients with advanced GC. Genomic information is now frequently employed for precision cancer treatment due to the rapid advancements of high-throughput sequencing technologies. As a result, integrating multiomics data to construct predictive models for the GC patient prognosis is critical for tailored medical care. Results: In this study, we integrated multiomics data to design a biological pathway-based gastric cancer sparse deep neural network (GCS-Net) by modifying the P-NET model for long-term survival prediction of GC. The GCS-Net showed higher accuracy (accuracy = 0.844), area under the curve (AUC = 0.807), and F1 score (F1 = 0.913) than traditional machine learning models. Furthermore, the GCS-Net not only enables accurate patient survival prognosis but also provides model interpretability capabilities lacking in most traditional deep neural networks to describe the complex biological process of prognosis. The GCS-Net suggested the importance of genes (UBE2C, JAK2, RAD21, CEP250, NUP210, PTPN1, CDC27, NINL, NUP188, and PLK4) and biological pathways (Mitotic Anaphase, Resolution of Sister Chromatid Cohesion, and SUMO E3 ligases) to GC, which is consistent with the results revealed in biological- and medical-related studies of GC. Conclusion: The GCS-Net is an interpretable deep neural network built using biological pathway information whose structure represents a nonlinear hierarchical representation of genes and biological pathways. It can not only accurately predict the prognosis of GC patients but also suggest the importance of genes and biological pathways. The GCS-Net opens up new avenues for biological research and could be adapted for other cancer prediction and discovery activities as well.

A classification method of gastric cancer subtype based on residual graph convolution network.

Background: Clinical diagnosis and treatment of tumors are greatly complicated by their heterogeneity, and the subtype classification of cancer frequently plays a significant role in the subsequent treatment of tumors. Presently, the majority of studies rely far too heavily on gene expression data, omitting the enormous power of multi-omics fusion data and the potential for patient similarities. Method: In this study, we created a gastric cancer subtype classification model called RRGCN based on residual graph convolutional network (GCN) using multi-omics fusion data and patient similarity network. Given the multi-omics data's high dimensionality, we built an artificial neural network Autoencoder (AE) to reduce the dimensionality of the data and extract hidden layer features. The model is then built using the feature data. In addition, we computed the correlation between patients using the Pearson correlation coefficient, and this relationship between patients forms the edge of the graph structure. Four graph convolutional network layers and two residual networks with skip connections make up RRGCN, which reduces the amount of information lost during transmission between layers and prevents model degradation. Results: The results show that RRGCN significantly outperforms other classification methods with an accuracy as high as 0.87 when compared to four other traditional machine learning methods and deep learning models. Conclusion: In terms of subtype classification, RRGCN excels in all areas and has the potential to offer fresh perspectives on disease mechanisms and disease progression. It has the potential to be used for a broader range of disorders and to aid in clinical diagnosis.

In vitro and in vivo studies of biodegradable Zn-Li-Mn alloy staples designed for gastrointestinal anastomosis.

Zn-0.8 wt.% Li-0.1 wt.% Mn wire with the diameter of 0.3 mm was fabricated and further processed into gastrointestinal staple, and its in vitro and in vivo biodegradation behavior and biocompatibility were studied systematically. The experimental Zn-Li-Mn alloy staple could deform from the original U-shape to B-shape without fracture, indicating its good mechanical property. Due to the residual stress concentration caused by anastomosis deformation, the feet and leg arc part of the staple were more prone to degradation. The Zn-Li-Mn alloy staple sustained integrity after immersion in Hanks' solution and simulated gastric fluid (SGF) for 28 days, and the degradation rate in SGF was about 4 times of that in Hanks' solution. Furthermore, Zn-Li-Mn alloy staples were utilized for gastrointestinal anastomosis in pig models, with clinically-used titanium alloy staples as a comparison. No anastomotic leakage and severe inflammation were observed after operation. The Zn-Li-Mn alloy staple maintained mechanical integrity within 8 weeks' implantation. The gastrointestinal tissue healed after 12 weeks, and no obvious side effects were detected during the whole implantation period, demonstrating the good biocompatibility of Zn-Li-Mn alloy staple. Thus, Zn-Li-Mn alloy staple fabricated in this work displayed the promising potential in the gastrointestinal anastomosis.

Comprehensive Analysis of CDCAs Methylation and Immune Infiltrates in Hepatocellular Carcinoma.

BACKGROUND: As essential components of cycle growth, the cell division cycle-associated family genes (CDCAs) have crucial roles in tumor development and progression, especially in hepatocellular carcinoma (HCC). However, due to the tumor heterogeneity of HCC, little is known about the methylation variability of CDCAs in mediating phenotypic changes (e.g., immune infiltrates) in HCC. Presently, we aim to comprehensively explore the expression and prognosis of CDCAs methylation with regard to immune infiltrates of HCC. METHODS: We first identified the correlating differentially expressed genes (co-DEGs) among 19 different types of cancer cohorts (a total of 7,783 patients) and then constructed the weighted gene co-expressed and co-methylated networks. Applying the clustering analysis, significant modules of DEGs including CDCAs were selected and their functional bioinformatics analyses were performed. Besides, using DiseaseMeth and TIMER, the correlation between the methylation levels of CDCAs and tumor immune infiltrates was also analyzed. In final, to assess the influence of CDCAs methylation on clinical prognosis, Kaplan-Meier and Cox regression analysis were carried out. RESULT: A total of 473 co-DEGs are successfully identified, while seven genes of CDCAs (CDCA1-3 and CDCA5-8) have significant over-expression in HCC. Co-expressed and co-methylated networks reveal the strong positive correlations in mRNA expression and methylation levels of CDCAs. Besides, the biological enrichment analysis of CDCAs demonstrates that they are significantly related to the immune function regulation of infiltrating immune cells in HCC. Also, the methylation analysis of CDCAs depicts the strong association with the tumor immunogenicity, i.e., low-methylation of CDCA1, CDCA2, and CDCA8 dramatically reduced the immune infiltrate levels of T cells and cytotoxic lymphocytes. Additionally, CDCA1-6 and CDCA8 with low-methylation levels significantly deteriorate the overall survival of patients in HCC. CONCLUSIONS: The co-expressed and co-methylated gene networks of CDCAs show a powerful association with immune function regulation. And the methylation levels of CDCAs suggesting the prognostic value and infiltrating immune differences could be a novel and predictive biomarker for the response of immunotherapy.

Programmed Death Receptor 1 (PD1) Knockout and Human Telomerase Reverse Transcriptase (hTERT) Transduction Can Enhance Persistence and Antitumor Efficacy of Cytokine-Induced Killer Cells Against Hepatocellular Carcinoma.

BACKGROUND The weak antitumor efficacy and limited lifespan are the main obstacles that hinder the therapeutic effect of cytokine-induced killer (CIK) cell immunotherapy. In the study, we enhanced the persistence and the antitumor efficacy of CIK cell through PD-1 knockout and hTERT transduction. MATERIAL AND METHODS CIK cells were cultured from patients with hepatocellular carcinoma and PD-1 gene was knocked out through the Cas9 ribonucleoproteins (Cas9 RNPs) electroporation. TIDE assay, T7E1 mismatch cleavage assay, and clone Sanger sequencing were used to detect PD-1 knockout efficiency. The immunophenotype was analyzed by flow cytometry. After PD-1 knockout, the hTERT gene was transduced into PD-1 KO/CIK cells with lentiviral transduction. The hTERT expression and persistence of hTERT/PD-1 KO/CIK cells were evaluated by Western blotting and proliferation curve. The antitumor efficacy was detected by ELISPOT and cytotoxicity assay. The telomere length was measured by the Q-FISH and qPCR method. The karyotype assay was used to analyze the chromosome structural stability. RESULTS The optimal knockout efficiency of PD-1 gene in CIK cells could reach 41.23±0.52%. PD-1 knockout did not affect the immunophenotype of CIK cells. The hTERT transduction enhanced persistence and increased the telomere length. ELISPOT and cytotoxicity assay showed hTERT/PD-1 KO/CIK cells had an enhanced antitumor efficacy. Meanwhile, PD-1 KO/CIK cells transduced with hTERT showed a normal karyotype. CONCLUSIONS PD-1 knockout combined with hTERT transduction could prolong the lifespan and enhance antitumor efficacy of CIK cells against hepatocellular carcinoma cell line.

[Sequence analysis of cloned duck hepatitis B virus genome from a Chongqing brown duck].

OBJECTIVE: To clone and analyze duck hepatitis B virus genome from Chongqing brown duck. METHODS: Duck hepatitis B virus (DHBV) DNA extracted from a Chongqing brown duck was amplified by PCR and cloned into PGEM-T vector using T-A clone method. The sequence of this DHBV genome was analyzed with some softwares after identified. RESULTS: The duck hepatitis B virus genome from Chongqing brown duck (DHBVcq), which was 3 024 nucleotides long, contained three ORFs whose onset and end nucleotides were in accord with those of HPUGA, encoding P, PreC/C and PreS/S protein respectively. Comparison of this strain with other DHBV reported in GenBank showed that the homology of DHBVcq and M32990 got the highest score of 94.9% at nucleotide level, while DHBVcq and DHBVCG got the least (89.8%). Most of the conserved regulation nucleotides and amino acids sequence found in other DHBV were also identified in DHBVcq. The epsilon region of DHBVcq, which was important for encapsidation of pgRNA and synthesis of minus-strand DNA, differed from that of most other DHBV strains, forming a stem-loop conformation with a three- nucleotides upper stem rather than a common nine-nucleotides one in free status. CONCLUSION: The successful clone and analysis of DHBVcq provide further studies with helpful information.