OBJECTIVE To construct a preoperative medication reconciliation model assisted by a localized large language model （LLM） for gastric cancer and evaluate its clinical efficacy. METHODS A total of 249 gastric cancer patients with a history of continuous medication before admission in the Gastric Surgery Department of Jiangsu Cancer Hospital were retrospectively enrolled. Patients were divided into training set （154 cases） and validation set （95 cases） based on the order of time. Based on guidelines， drug package inserts， and other evidence， a standardized medication reconcili ation process and a structured knowledge base were constructed. DeepSeek-V3 LLM was deployed privately in the hospital， combined with retrieval-augmented generation technology， to achieve automated integration of medication information， risk screening， and generation of personalized recommendations. The quality of LLM-generated recommendations was evaluated using automatic metrics （BERT Score and ROUGE-1， 2， L） and manual scoring [seven-dimensional index （7DI） ] . Spearman correlation analysis was performed to explore the correlation between automatic scores and manual scores. Cronbach’s α coefficient was used to test the internal consistency of manual scoring results. The time consumed by manual and LLM-assisted medication reconciliation was compared across tasks of different difficulty levels （simple， moderate， and high）. RESULTS A structured knowledge base covering 8 major drug categories was finally established， covering common and high-risk preoperative medication scenarios and providing structured retrieval support for the LLM. For automatic evaluation， the precision， recall， and F1-score of BERT Score were 0.783±0.033， 0.811±0.038， and 0.796±0.028， respectively. The F1-scores of ROUGE-1， ROUGE-2 and ROUGE-L were 0.566±0.067， 0.338±0.076 and 0.468±0.082， respectively. The 7DI scores from three manual raters ranged from 32.06 to 33.45. The F1-score of automatic scoring was significantly positively correlated with the 7DI score of manual scoring （maximum coefficient of determination＝0.611， P ＜0.001）， and the internal consistency of manual scoring was good （Cronbach’s α ＝ 0.876）. In terms of efficiency， LLM-assisted medication reconciliation reduced time consumption by more than 90% compared with manual reconciliation in the simple， moderate， and high-difficulty groups （ P ＜0.001）. CONCLUSIONS The medication reconciliation model constructed based on a localized LLM and structured knowledge base shows high accuracy， consistency， and clinical applicability in complex preoperative medication scenarios for gastric cancer. It can improve the efficiency of medication reconciliation and reduce potential medication risks.

Objective: To conduct network analysis of biased constitution to provide theoretical basis for daily health management. Methods: Purposive sampling and snowball sampling methods were used to collect the required data via a web-based questionnaire from February 3rd, 2022 to January 21st, 2023. The tendency scores of 8 biased constitutions (qi-deficiency type, yang-deficiency type, yin-deficiency type, phlegm-dampness type, damp-heat type, blood-stasis type, qi-depression type, and special diathesis type) were evaluated by 30-Item Short Version of Constitution in Chinese Medicine Questionnaire. Data analysis was conducted in R. The qgraph package was used to construct network analysis models and calculate the edge weight, namely regularised partial correlation coefficients (r) and centrality coefficients (including strength, closeness, betweenness and expected influence), while mgm was used to calculate the predictability of each node within the network. The ggplot2 package was used to visualize network analysis result, and bootnet was applied to assess the structural reliability and robustness of the network, including the r accuracy assessment, the stability assessment of the centrality coefficients, and the difference test result of node centrality coefficients. Results: A total of 3, 691 valid samples were obtained from the survey. The result of the network analysis showed that: (1) the strength of the association between phlegm dampness type and damp-heat type (r=0.370), qi-deficiency type and qi-depression type (r =0.315), qi-deficiency type and special diathesis type (r=0.260), qi-deficiency type and phlegm-dampness type (r=0.247), and the phlegm-dampness type and qi-depression type (r=0.247) was high. (2) The centrality of phlegm-dampness type and qi-deficiency type was the strongest, and the differences between the centrality coefficients of phlegm-dampness type and qi-deficiency type were not significant. There were no other nodes where all 4 centrality coefficients were not significantly different from qi-deficiency type and phlegm-dampness type, and the predictability of phlegm-dampness type and qi-deficiency type was the highest. (3)The result of the r accuracy assessment showed that the r of this network did not differ significantly from the average of multiple samples, so the r accuracy of this network was high. (4) Correlation stability(CS) coefficient of the four centrality indicators were CS Strength=0.59, CS Closeness=0.75, CS Betweenness=0.75, and CS Expected influence=0.75. This indicated that the stability of the centrality indicators of this network was very good. Conclusion Complex connections exist between various biased constitutions, and phlegm-dampness type and qi-deficiency type are the two key biased constitutions in the entire biased constitution.

Objective:Based on bioinformatics analysis, this study aimed to identify the complement component 5a receptor 1(C5aR1)and chemokine C-C motif ligand-2(CCL2)genes in vascular dementia(VaD)and to explore the expression and clinical significance of serum C5aR1 and CCL2 levels in VaD patients.Methods:The GSE122063 dataset was selected from the Gene Expression Omnibus(GEO)database to screen for consistently differentially expressed genes in the frontal and temporal lobes of VaD patients and non-dementia patients.The Matascape database was used to analyze the functions and pathways of differentially expressed genes, and the STRING network and Cytoscape software were used to identify key genes.In this case-control study, 53 VaD patients seeking care at the Department of Neurology of the Affiliated Hospital of Jiangsu University between January 2022 and December 2022 were included in the VaD group, and 50 non-dementia individuals were included in the control group.General information, Montreal Cognitive Assessment(MoCA)scores, Mini-Mental State Examination(MMSE)scores, and scores of the total cerebral small vessel disease(CSVD)burden were collected for both groups, and serum C5aR1 and CCL2 expression was detected.The correlation of serum C5aR1 and CCL2 levels with MoCA scores, MMSE scores, and scores of the total CSVD burden in the VaD group was analyzed.Receiver operating characteristic(ROC)curve analysis was used to assess the diagnostic value of serum C5aR1 and CCL2 levels in VaD.Results:In the GSE122063 dataset, compared with non-dementia patients, there were 43 upregulated genes and 63 downregulated genes simultaneously in the frontal and temporal lobes in the VaD group.After importing 106 genes into the Cytoscape software and using the Stress and Betweenness algorithms in the cytoHubba plugin, two key genes, C5aR1 and CCL2, were identified.Serum levels of C5aR1[(57.25±10.34)μg/L vs.(43.26±8.24)μg/L, t=7.607, P<0.001]and CCL2[(210.42±42.19)ng/L vs.(151.73±36.04)ng/L, t=7.570, P<0.001]in the VaD group were higher than those in the control group.Serum levels of C5aR1 and CCL2 were negatively correlated with MoCA scores( r=-0.691, -0.668, P<0.001)and MMSE scores( r=-0.736, -0.729, P<0.001), and positively correlated with scores of the total CSVD burden( r=0.598, 0.582, P<0.001).The areas under the ROC curve for serum C5aR1 and CCL2 levels in diagnosing VaD was 0.838 and 0.845, respectively.The area under the ROC curve with the combination of C5aR1 and CCL2 for the diagnosis of VaD was 0.896. Conclusions:Serum levels of C5aR1 and CCL2 are elevated in VaD patients and closely related to their cognitive function and the total CSVD burden, and may be used as an auxiliary diagnostic tool for VaD patients.

The incidence of hypertriglyceridemic acute pancreatitis(HTG-AP)has been gradually rising in recent years.According to published findings,high triglyceride levels are strongly associated with the severity of acute pancreatitis,and may lead to a higher incidence of complications and worse prognosis.However,the risk factors associated with HTG-AP have not been systematically explored.Early identification and effective management of high triglyceride levels and other potential causes thereof are crucial for reducing the recurrence of acute pancreatitis in clinical practice.Herein,we reviewed the predictive factors of HTG-AP from the perspectives of etiology,pathogenesis,and the relevant biomarkers,such as C-reactive protein,Ca2+,PT,and D-dimer.We aim to provide important early warning signals for clinicians,thereby helping develop personalized treatment protocols and building a more accurate risk prediction model.

Objective:To analyze the current situation of health-related quality of life (HRQOL) and its influencing factors among community-dwelling elderly with mild cognitive impairment (MCI) based on the health ecology model (HEM).Methods:From December 2023 to September 2024, a cross-sectional survey was conducted among 997 community-dwelling elderly in Beijing, Shandong, Chongqing and other regions. The EuroQol five-dimensional questionnaire 5 level (EQ-5D-5L), the Montreal cognitive assessment (MoCA), and the Pittsburgh sleep quality index (PSQI) were used to assess the subjects.Descriptive analysis, Mann-Whitney U test, Kruskal-Wallis test, and Tobit regression analysis were performed using Stata 18.0 software. Results:A total of 324 valid samples were finally obtained, and the health utility value of MCI elderly was 0.842(0.815, 0.951). Tobit regression analysis showed that ≥80 years old( β=-0.121, 95% CI=-0.193--0.050) in individual traits level, don't exercise( β=-0.196, 95% CI=-0.255--0.137) and poor quality of sleep ( β=-0.064, 95% CI=-0.116--0.013) in behavior characteristics level, and poor subjective economic status( β=-0.153, 95% CI=-0.261--0.045) in living and working conditions level were risk factors for the health utility value of the elderly with MCI. Drinking( β=0.096, 95% CI=0.022-0.171) in behavior characteristics level and participating in social activities( β=0.126, 95% CI=0.062-0.190) in interpersonal network level were protective factors for the health utility value of the elderly with MCI. Conclusions:The HRQOL of the community-dwelling elderly with MCI in China is low, and its influencing factors are multi-level. HEM should be combined to strengthen the intervention and management of the elderly with MCI from personal constitution to policy environment to improve their HRQOL.

Phage immunoprecipitation sequencing (PhIP-Seq) is a high-throughput and low-cost method for analyzing the specific binding of target proteins to peptide libraries. The method uses oligonucleotide library synthesis (OLS) to encode proteome-scale peptide libraries for display on phages, and then immunoprecipitates these library phages with target proteins (such as antibodies) for subsequent analysis by high-throughput DNA sequencing. PhIP-Seq enables the screening of peptide targets that react specifically with hundreds of proteins or pathogens. PhIP-Seq has been successfully applied in various fields such as disease detection, screening of autoimmune disease biomarkers, vaccine development, and allergen detection, becoming a high-throughput diagnostic technology. This article systematically describes the development, applications, and result evaluation of PhIP-Seq, in order to gain a more comprehensive understanding of the application and future development prospects of this technology in various fields.
Peptide Library ; Humans ; Immunoprecipitation/methods* ; High-Throughput Nucleotide Sequencing/methods* ; Bacteriophages/genetics*

Objective:To investigate the effect of hypericin (Hyp) on neurologic impairment, ferroptosis and cuproptosis in mice with acute cerebral infarction.Methods:Sixty 8-week old male C57BL/6 mice were randomly divided into sham-operated group, middle cerebral artery occlusion (MCAO) group, MCAO+Hyp low-dose treatment group (L-Hyp group), and MCAO+Hyp high-dose treatment group (H-Hyp group), with 15 mice in each group. Intraluminal filament MCAO models in the later 3 groups were established. Saline was given intraperitoneally into the sham-operated group and MCAO group, and Hyp was given intraperitoneally at 0.5 mg/kg or 1 mg/kg into the L-Hyp group and H-Hyp group 24 hours after modeling. Twenty-four hours after Hyp, neurologic function was assessed using Garcia score, grip strength test, and fatigue baton test; brain tissue edema was assessed by dry-wet weight method; neuronal necrosis, survival and apoptosis were detected by HE staining, Nissl staining and TUNEL, respectively; ferroptosis and oxidative stress were assessed using iron assay kit, and reactive oxygen species (ROS), malondialdehyde (MDA) and glutathione (GSH) assay kits; cuproptosis was assessed using copper assay kit and mitochondrial oxidative phosphorylation was evaluated by mitochondrial respiratory chain complex Ⅲ and Ⅳ activity detection kits; morphological changes in neuronal mitochondria after ferroptosis and cuproptosis were observed by electron microscopy; protein expressions of ferroptosis-associated solute carrier family 7 member 11 (SLC7A11), glutathione peroxidase 4 (GPX4), cuproptosis-associated solute carrier family 31 member 1 (SLC31A1), and ferredoxin 1 (FDX1) were detected by Western blotting.Results:(1) Compared with the MCAO group, the L-Hyp group and H-Hyp group had decreased modified Garcia score and brain water content, increased grip strength and rod-turning time, decreased number of necrotic and apoptotic neurons, increased number of survived neurons, decreased Fe 2+, ROS and MDA levels, increased GSH level and mitochondrial respiratory control rate, and decreased copper content, with significant differences ( P<0.05); and the above changes in the H-Hyp group were more obvious than those in the L-Hyp group, with significant differences ( P<0.05). Compared with the MCAO group, neurons in the L-Hyp group and H-Hyp group had significantly improved status in mitochondrial shrinkage, vacuolation, reduced cristulation, increased membrane density, ruptured cell membrane, endoplasmic reticulum damage and chromatin disruption ( P<0.05); and the H-Hyp group had signficantly more obvious improvement than the L-Hyp group ( P<0.05). (2) Compared with the MCAO group (0.38±0.09, 0.28±0.05), the L-Hyp group and H-Hyp group had significantly increased protein expressions of SLC7A11 and GPX4 (0.83±0.11, 0.49±0.06; 1.27±0.08, 0.84±0.04; P<0.05); the H-Hyp group had significantly higher SLC7A11 and GPX4 expressions than the L-Hyp group ( P<0.05). Compared with the MCAO group (2.76±0.17, 0.67±0.07), the L-Hyp group and H-Hyp group had significantly decreased protein expressions of SLC31A1 and FDX1 (1.72±0.07, 0.51±0.05; 1.12±0.06, 0.34±0.05; P<0.05); the H-Hyp group had significantly lower SLC31A1 and FDX1 expressions than the L-Hyp group ( P<0.05). Conclusion:Hyp can ameliorate ferroptosis and cuproptosis by regulating the protein expressions of SLC7A11, GPX4, SLC31A1 and FDX1, to alleviate oxidative stress injury in MCAO mice, thereby promoting the recovery of neurological function.

Objective To investigate the predictive value of edema metabolic and hematoma dynamics changes on motor and cog-nitive recovery outcomes in patients with intracerebral hemorrhage(ICH).Methods The CT data of ICH patients were collected to evaluate hematoma volume changes from admission to day 3.On day 3,multivoxel magnetic resonance spectroscopy(MRS)was per-formed with region of interest located in the edema region and contralateral normal tissue.Motor and cognitive function recovery was assessed using the simplified F-M scale and the Montreal cognitive assessment(MoCA)on day 3 and at the 3-month follow-up,respec-tively.Overall clinical outcomes were assessed using the Glasgow outcome scale(GOS),and all patients were divided into good and poor outcome groups.Clinical data and metabolic differences in the edema region between the two groups were compared,respec-tively.Logistic regression analysis and receiver operating characteristic(ROC)curves were used to identify and evaluate independent prognostic factors.Subgroup analysis were performed via stratification of hematoma location.Results The logistic regression analy-sis indicated that intraventricular extension,hematoma changes,and the ratio of N-acetyl aspartate(NAA)around the hematoma to contralateral normal brain parenchyma NAA(rNAA)were inde-pendent prognostic factors for poor outcomes(P＜0.05).The area under the curve(AUC)for each factor and the combined model were 0.69,0.73,0.79,and 0.82,respectively.In patients with ICH in the basal ganglia region,△F-M was negatively correlated with hematoma changes and positively correlated with rNAA value(P＜0.001).In patients with ICH in the thalamic and lobar regions,△MoCA was not significantly correlated with hematoma changes(P＞0.05),but was positively correlated with rNAA value(P＜0.001).Conclusion The rNAA holds predictive value for motor and cognitive recovery outcomes following standard treatment.

Objective:To evaluate the utility of ultrasonographic monitoring of the rectus femoris muscle—specifically, the rates of change in thickness and cross-sectional area (CSA)—in assessing nutritional status and long-term functional outcomes in patients with sepsis.Methods:In this prospective observational study, sepsis patients admitted to the ICU of the Second Affiliated Hospital of Soochow University between October 2023 and October 2024 were classified by nutritional status at discharge using the Global Leadership Initiative on Malnutrition (GLIM) criteria. Differences in serial ultrasound-measured rectus femoris thickness and CSA on days 1, 3, 5, and 7 were compared between malnourished and non-malnourished groups. The predictive value of these ultrasound parameters for malnutrition was analyzed. Functional prognosis was assessed using the Sarcopenia Assessment Scale, Short Physical Performance Battery, and Manual Muscle Testing, with correlations to muscle changes examined.Results:Of the 71 enrolled patients (median age 73.00 [ IQR: 61.00–80.00]; 47.89% female, 52.11% male), those with malnutrition showed significantly greater variation rates in rectus femoris thickness and CSA on days 3, 5, and 7 compared to the non-malnourished group ( P < 0.05). ROC analysis revealed that the day-7 CSA variation rate had the highest predictive value for malnutrition (AUC = 0.817, 95% CI: 0.713-0.930). These muscle variation rates also correlated strongly with conventional nutritional markers such as BMI, albumin, and urea. Similarly, patients with impaired functional outcomes exhibited higher variation rates in muscle parameters on days 3, 5, and 7 ( P < 0.05), with the day-7 CSA variation rate being most predictive of functional prognosis (AUC = 0.749, 95% CI: 0.632-0.867). Conclusions:Ultrasonographic assessment of rectus femoris thickness and CSA variation rates provides a valuable tool for evaluating nutritional status and predicting functional prognosis in sepsis patients, outperforming traditional biomarkers. This method shows promise for guiding individualized nutrition support and rehabilitation strategies to improve long-term outcomes.

Objective:To analyze the changes of serum metabolites in patients with acute pancreatitis (AP) by non-targeted metabolomics method.Methods:Serum samples and clinical data of 15 AP patients hospitalized in the Affiliated Hospital of Jiangnan University from August to September 2024 were collected and included in the AP group, including 9 males and 6 females, aged (55.4±15.3) years. The serum and clinical data of 25 patients with colon polyps in the same hospital during the same period of time were collected, including 15 males and 10 females, aged (61.2±11.5) years, and were included in the control group. Serum metabolomic detection was performed using the ultra-high performance liquid chromatography tandem Fourier transform mass spectrometer. The modeling method was orthogonal partial least square discriminant analysis, and principal component analysis was performed on the data matrix to screen the differential metabolites in serum of AP patients. The Kyoto Encyclopedia database of Genes and Genomes was used to annotate differential metabolites, and the pathway of differential metabolite enrichment was analyzed by software.Results:The principal component analysis showed that the contribution ratio of the first principal component was 15.1%, the proportion of the second principal component was 10.8%, and the total proportion of the two was 25.9%. In principal component analysis, two groups of samples can be clearly distinguished and show obvious clustering characteristics. According to the analysis of OPLS-DA model, there were significant differences in serum metabolic profiles between AP group and control group. There were 683 differentially expressed metabolites between the two groups, with 367 differentially expressed metabolites up-regulated compared with the control group and 316 differentially expressed metabolites down-regulated compared with the control group. It is mainly Phosphatidic Acid (Lte4/8: 0) (+ 218%), Omeprazole Sulphone (-38%), and 2-(Propylthio) Nicotinic Acid (2-propyl thionicotinic acid) (-58%), Gein (salicyricetin) (-47%) and so on. Pathway enrichment analysis showed that the differential metabolites in AP patients were mainly concentrated in citric acid cycle, arginine biosynthesis and glycerophospholipid metabolism pathways.Conclusion:Serum metabolites in AP patients change significantly, including citric acid cycle, arginine biosynthesis, glycerophospholipid metabolism.