ObjectiveThe essence of syndrome manifestation recognition in traditional Chinese medicine (TCM) is to infer the body’s latent pathogenesis state from clinical observational information, rather than to perform simple label matching. However, previous studies have largely modeled this task as syndrome pattern classification within a fixed label space, which does not adequately reflect the cognition process of TCM syndrome differentiation centered on pathogenesis reasoning, and is also insufficient to capture the openness, semantic variability, and cross-disease reusability of syndrome manifestation expression. This study aimed to investigate whether introducing pathogenesis reasoning chain-of-thought (PR-CoT) supervision into large language models (LLMs) could improve the quality and cognitive consistency of syndrome manifestation recognition and support cross-disease transfer. MethodsSyndrome manifestation recognition was formulated as a conditional generation task under the framework of clinical observational information (X)→pathogenesis structure (Z)→syndrome pattern output (Y), where Z serves as an explicit intermediate structural variable linking the clinical evidence and syndrome judgment. Within this framework, a PR-CoT-supervised dataset for syndrome manifestation recognition was constructed based on medical case records of spleen-stomach disorders. After preprocessing, information extraction, manual proofreading, and data cleaning, the dataset comprised 4 800 training cases, 400 development cases, and 400 test cases. Each sample was annotated with a structured PR-CoT consisting of three progressive levels: clinical information summarization, comprehensive pathogenesis analysis, and syndrome pattern output. Supervised fine-tuning was conducted on open-source LLMs, with an end-to-end model serving as the baseline. Qwen3-32B was used as the primary experimental model, and Qwen3-14B as the scale comparison model. A progressive multidimensional evaluation framework was further established, comprising a structural parsing level, a semantic similarity level, and an expert blind review level. At the structural parsing level, syndrome pattern expressions were decomposed into structural elements and evaluated using Precision, Recall, F1 score, and Jaccard similarity. At the semantic similarity level, independent LLMs scored the theoretical proximity between predicted and reference syndrome patterns. At the expert blind review level, three TCM experts independently evaluated model outputs on two dimensions: syndrome differentiation consistency and terminology standardization of syndrome patterns. In addition, zero-shot cross-disease transfer evaluation was conducted on gynecological and heart-system disorder test sets. ResultsAt the structural parsing level, PR-CoT supervision did not lead to a stable improvement in the element-wise overlap of syndrome pattern structural components. Compared with the corresponding baselines, neither Qwen3-32B nor Qwen3-14B showed consistent advantages in structural matching metrics after the introduction of PR-CoT supervision. In contrast, at the semantic similarity level, PR-CoT supervision produced stable positive gains across different model scales and evaluation systems. The average semantic score of Qwen3-32B increased from 6.425 8 in the baseline model to 6.585 0 after PR-CoT supervision, and that of Qwen3-14B increased from 5.870 0 to 5.964 2. At the expert blind review level, the overall score of Qwen3-32B (PR-CoT) was 7.026 0±0.107 7, higher than 6.416 3±0.288 9 for its baseline. In zero-shot cross-disease testing, the PR-CoT model still showed advantages in semantic evaluation and expert evaluation on both gynecological and heart-system disorder test sets, indicating a certain degree of transferability. ConclusionThe benefits of PR-CoT supervision are mainly reflected in TCM semantic consistency and clinical plausibility, rather than in improved hard matching of structural elements. These findings support understanding syndrome manifestation recognition as a process of generating and expressing latent pathogenesis structures, rather than as a classification task within a traditional fixed label space. By introducing pathogenesis reasoning as an explicit intermediate structure into the modeling process and combining it with a progressive multidimensional evaluation framework, this study provides a methodological pathway for intelligent TCM syndrome differentiation that integrates theoretical alignment, interpretability, and multi-level evaluation.

ObjectiveThe essence of syndrome manifestation recognition in traditional Chinese medicine (TCM) is to infer the body’s latent pathogenesis state from clinical observational information, rather than to perform simple label matching. However, previous studies have largely modeled this task as syndrome pattern classification within a fixed label space, which does not adequately reflect the cognition process of TCM syndrome differentiation centered on pathogenesis reasoning, and is also insufficient to capture the openness, semantic variability, and cross-disease reusability of syndrome manifestation expression. This study aimed to investigate whether introducing pathogenesis reasoning chain-of-thought (PR-CoT) supervision into large language models (LLMs) could improve the quality and cognitive consistency of syndrome manifestation recognition and support cross-disease transfer. MethodsSyndrome manifestation recognition was formulated as a conditional generation task under the framework of clinical observational information (X)→pathogenesis structure (Z)→syndrome pattern output (Y), where Z serves as an explicit intermediate structural variable linking the clinical evidence and syndrome judgment. Within this framework, a PR-CoT-supervised dataset for syndrome manifestation recognition was constructed based on medical case records of spleen-stomach disorders. After preprocessing, information extraction, manual proofreading, and data cleaning, the dataset comprised 4 800 training cases, 400 development cases, and 400 test cases. Each sample was annotated with a structured PR-CoT consisting of three progressive levels: clinical information summarization, comprehensive pathogenesis analysis, and syndrome pattern output. Supervised fine-tuning was conducted on open-source LLMs, with an end-to-end model serving as the baseline. Qwen3-32B was used as the primary experimental model, and Qwen3-14B as the scale comparison model. A progressive multidimensional evaluation framework was further established, comprising a structural parsing level, a semantic similarity level, and an expert blind review level. At the structural parsing level, syndrome pattern expressions were decomposed into structural elements and evaluated using Precision, Recall, F1 score, and Jaccard similarity. At the semantic similarity level, independent LLMs scored the theoretical proximity between predicted and reference syndrome patterns. At the expert blind review level, three TCM experts independently evaluated model outputs on two dimensions: syndrome differentiation consistency and terminology standardization of syndrome patterns. In addition, zero-shot cross-disease transfer evaluation was conducted on gynecological and heart-system disorder test sets. ResultsAt the structural parsing level, PR-CoT supervision did not lead to a stable improvement in the element-wise overlap of syndrome pattern structural components. Compared with the corresponding baselines, neither Qwen3-32B nor Qwen3-14B showed consistent advantages in structural matching metrics after the introduction of PR-CoT supervision. In contrast, at the semantic similarity level, PR-CoT supervision produced stable positive gains across different model scales and evaluation systems. The average semantic score of Qwen3-32B increased from 6.425 8 in the baseline model to 6.585 0 after PR-CoT supervision, and that of Qwen3-14B increased from 5.870 0 to 5.964 2. At the expert blind review level, the overall score of Qwen3-32B (PR-CoT) was 7.026 0±0.107 7, higher than 6.416 3±0.288 9 for its baseline. In zero-shot cross-disease testing, the PR-CoT model still showed advantages in semantic evaluation and expert evaluation on both gynecological and heart-system disorder test sets, indicating a certain degree of transferability. ConclusionThe benefits of PR-CoT supervision are mainly reflected in TCM semantic consistency and clinical plausibility, rather than in improved hard matching of structural elements. These findings support understanding syndrome manifestation recognition as a process of generating and expressing latent pathogenesis structures, rather than as a classification task within a traditional fixed label space. By introducing pathogenesis reasoning as an explicit intermediate structure into the modeling process and combining it with a progressive multidimensional evaluation framework, this study provides a methodological pathway for intelligent TCM syndrome differentiation that integrates theoretical alignment, interpretability, and multi-level evaluation.

Objective To analyze the epidemiological characteristics and disease burden of liver cancer in Guangdong Province in 2020, and to provide a scientific foundation for the development of regionalized prevention and control strategies for liver cancer. Methods According to the cancer registry data of Guangdong Province, the incidence, mortality and age-standardized rate by Chinese standard population in 2020 were calculated to analyze the epidemiological characteristics of liver cancer. The disability adjusted life years (DALYs), year of life loss (YLL), year of lived with disability (YLD), and cause-eliminated life expectancy were used to assess the disease burden of liver cancer. Results In 2020, the crude incidence rate and the age-standardized incidence rate of liver cancer in Guangdong Province were 27.79/100 000 and 20.84/100 000，respectively, and the crude mortality rate and the age-standardized mortality rate of liver cancer were 25.49/100,000 and 17.64/100 000, respectively. The total DALY and DALY rate of liver cancer in Guangdong Province were 515 311 person-years and 513.83/100 000, respectively. After eliminating the causes of death from liver cancer, the life expectancy in Guangdong Province increased from 84.60 years to 84.99 years. All indicators consistently demonstrated that the burden of liver cancer was higher in males than that in females, and the burden of liver cancer was higher in rural areas than that in urban areas. Conclusion Liver cancer in Guangdong Province exhibits a high incidence, mortality and disease burden level in 2020. There are obvious differences of gender, age and region in cancer burden. It is necessary to strengthen liver cancer screening and diagnosis and treatment in men, the elderly and those in rural areas to reduce the burden of liver cancer gradually in Guangdong Province.

Objective To evaluate the cost–utility of lecanemab and donanemab in the treatment of early Alzheimer’s disease （AD） in China. Methods A Markov model was developed from a societal perspective to compare standard of care （SOC）, lecanemab plus SOC, and donanemab plus SOC in terms of costs, quality-adjusted life years （QALY）, and incremental cost–utility ratios （ICUR）. Deterministic and probabilistic sensitivity analyses were conducted to assess parameter uncertainty. Results Both intervention strategies increased QALY compared with SOC but were associated with substantially higher costs, and neither was cost-effective under the current willingness-to-pay threshold. Conclusion Lecanemab and donanemab may be more appropriately considered within a multi-tiered healthcare reimbursement framework in China rather than relying solely on basic medical insurance.

Objective:To explore the interrelationship between brain functional networks and features in functional magnetic resonance imaging(fMRI)of patients with Alzheimer's disease(AD),and to construct mixed-function networks(MFN),and apply them in machine learning classification models to improve the accuracy of AD classification.Methods:102 AD patients and 227 healthy subjects in the Alzheimer's Neuroimaging Initiative(ADNI)dataset were retrospectively analyzed.The partial correlation brain network of the blood oxygen level dependent(BOLD)signal was calculated and fused with low-frequency wave amplitude(ALFF),fractional low-frequency wave amplitude(fALFF)and local consistency(ReHo)features to construct MFN.Network topology parameters were extracted,and a variety of machine learning classification models were constructed based on MFN topological parameters,accuracy,precision,recall and area under the curve(AUC)were used to evaluate the predictive efficiency of the models.Results:By constructed MFN and calculated intra group to inter group ratio(IIGR),35 features could be obtained from ALFF,fALFF and ReHo feature topological parameter analysis,after rank sum test and FDR correction,there were statistical differences among 28 features(P＜0.05).The classification results show that,all the five classifiers have high classification performance on the test data set.The accuracy,precision and recall rates of random forest(RF),adaptive lifting algorithm(AdaBoost),guided aggregation algorithm(Bagging)and support vector machine(SVM)were all 99.7％,and the AUC values were up to 100％,99.5％,99.1％and 99.5％,respectively.The accuracy(98.5％),precision(98.5％),recall(98.5％),and AUC(99.1％)of the multi-layer perceptron(MLP)were slightly lower than other models,but remained excellent.It was worth noting that RF has the highest AUC value of all models at 100.0％,while Bagging has the lowest AUC value(99.1％)in the integrated approach.The results of performance comparison show that,MFN classification model can significantly improve the recognition and classification of AD disease,and greatly improve the performance of various indicators of the classifier.The results showed that,MFN classification model was superior to intelligent classification based fusion,DBN-based multitask learning,PVT-TSVM,unsupervised learning and clustering,SVM and SVM of degree 3 polynomial kernel function in key indicators such as accuracy(99.13％),AUC(99.42％),recall rate(99.46％)and specificity(99.42％)with plasma proteins,machine learning algorithms.It was further proved that MFN classification model has good generalization ability and robustness in AD disease classification.Conclusion:The AD classification model constructed based on brain mixed function network topology parameters and machine learning can improve the accuracy of AD classification.

Objective To explore the effectiveness of the OKR method in improving the efficiency of the whole process of critical value management,and to provide new ideas for the implementation of medical quality improvement in other medical insti-tutions.Methods A hospital in Tianjin was selected as the study object,which used the OKR method to reform the management of critical value since January 2024.The core goal of"improving the effectiveness of critical value management"was set,and the goal was broken down into quantitative key results such as"the rate of receiving the critical value system is over 95％"and"the rate of completing the standardised writing of medical records is up to 85％".Results After the reform,several quantitative key results,such as the completion rate of critical care medical record writing,the rate of standardised medical record writing,and the rate of overtime acceptance,were all better than before.Conclusion Through the OKR method to unify the whole hospital's strategic objectives,and the dynamic adjustment of the program based on data review,the hospital's critical value management efficiency has been significantly improved,effectively guaranteeing the safety of patients,and providing new perspectives and methods for the management of critical value in other medical institutions.

Purpose To explore the diagnostic value of two-dimensional speckle tracking echocardiography(2D-STE)in differentiating left ventricular hypertrophy(LVH)caused by Fabry disease from other etiologies.Materials and Methods A total of 23 patients clinically confirmed Fabry disease with LVH(Fabry disease group)in Peking University First Hospital from August 2014 to February 2023,retrospectively.23 patients with hypertensive LVH(hypertensive LVH group)and 23 with hypertrophic cardiomyopathy(HCM)(HCM group)were also included.Conventional echocardiographic parameters and 2D-STE-derived left atrial strain and left ventricular longitudinal strain were analyzed and compared among the three groups.LASSO regression was used to select variables and construct a diagnostic model to differentiate hypertensive LVH,HCM and Fabry disease group.Results The Fabry disease group showed significantly reduced left ventricular global longitudinal strain and segmental left ventricular longitudinal strain in the basal anterior wall,basal anterolateral wall and mid inferior wall compared to the hypertensive LVH group(P<0.05).There were no significant differences in left atrial strain,left ventricular global longitudinal strain and segmental left ventricular longitudinal strain between the Fabry disease and HCM groups(P>0.05).Compared to the conventional echocardiography-based model,the combined model did not significantly improve diagnostic accuracy for differentiating the three etiologies(Z=-1.314--0.594,all P>0.05).Conclusion Compared to patients with hypertensive LVH,HCM and Fabry disease group exhibit varying degrees of decrease in global and segmental longitudinal strain of the left atrium and left ventricle.Additional measurements of left ventricular longitudinal strain and left atrial strain do not provide significant incremental diagnostic value over conventional echocardiography in distinguishing among Fabry disease,hypertensive LVH and HCM.

Objective:To investigate the effects of combination therapy with aqueous extract of corn silk(CS)and training on exercise capacity and glycolipid metabolism in mice with metabolic syndrome(MS).Methods:In this study,db/db mice were used as the animal model of MS.The mice were administered aqueous extract of CS via gavage and subjected to different intensities of training for 12 weeks(3 months).The specific experimental design was as follows:24 db/db mice were randomly divided into four groups on average:negative control group(NC),aqueous extract of CS group(CS),aqueous extract of CS+moderate-intensity training group(CS+MT),and CS aqueous extract of CS+high-intensity training group(CS+HT).The maximum running speed,forelimb grip strength,body weight and fasting blood glucose of mice were measured before and after treatment.After the intervention,oral glucose tolerance test(OGTT)and insulin tolerance test(ITT)were conducted to assess glucose metabolism,while serum triglyceride(TG),total cholesterol(TC),high-density lipoprotein cholesterol(HDL-C),and low-density lipoprotein cholesterol(LDL-C)levels were measured to evaluate lipid metabolism.Results:After 3 months of intervention,there were significant differences in the maximum running speed and forelimb grip strength among the four groups(P＜0.05).The maximum running speed and forelimb grip strength of CS group,CS+MT group and CS+HT group were higher than those of NC group(P＜0.05).The CS+MT group exhibited higher forelimb grip strength,and the CS+HT group showed higher maximum running speed and forelimb grip strength compared to the CS group(P＜0.05),while no significant difference was found between the CS+MT and CS+HT groups(P＞0.05).Significant differences in body weight were observed among the four groups after 3 months of intervention(P＜0.05).Specifically,the CS+MT and CS+HT groups exhibited significantly lower body weight compared to both the NC and CS groups(P＜0.05),with the CS+MT group having the lowest body weight(P＜0.05).Fasting blood glucose levels also differed significantly among the groups after 2 and 3 months of intervention(P＜0.05).The CS,CS+MT,and CS+HT groups had lower fasting blood glucose levels compared to the NC group(P＜0.05),with the CS+MT and CS+HT groups showing the lowest levels(P＜0.05).No significant difference was found between the CS+MT and CS+HT groups(P＞0.05).After 3 months of intervention,significant differences in the area under the curve(AUC)of OGTT and ITT were observed among the four groups(P＜0.05).The AUC of OGTT and ITT were significantly lower in the CS,CS+MT,and CS+HT groups compared to the NC group(P＜0.05).The CS+MT and CS+HT groups exhibited the lowest AUC values for both OGTT and ITT(P＜0.05),with the CS+MT group showing the lowest AUC for OGTT(P＜0.05).Significant differences in serum lipid levels were observed among the four groups after 3 months of intervention(P＜0.05).TG,TC,and LDL-C levels were significantly lower,while HDL-C levels were higher in the CS,CS+MT,and CS+HT groups compared to the NC group(P＜0.05).The CS+MT group had the lowest TG levels and the highest HDL-C levels compared to the CS+HT group(P＜0.05),with no significant differences in TC and LDL-C levels between these two groups(P＞0.05).Conclusion:Aqueous extract of CS combined with different intensity training can significantly improve the exercise capacity and glycolipid metabolism of MS mice and reduce body weight,especially CS combined with MT treatment is more effective in improving lipid metabolism.In addition,when combined with HT,aqueous extract of CS can also play an auxiliary role in reducing the side effects of high-intensity exercise and improving the therapeutic effect.

Objective To compare the value of ultrashort echo time(UTE)and T1W sequences for detecting bone erosions of gouty arthritis.Methods Forty-four gouty patients were prospectively enrolled,including 32 cases with affected feet and 12 cases with affected knee.MR UTE and T1W sequence scanning of the affected area were performed,and subjectively scoring of imaging quality of 2 kinds of MRI were evaluated,respectively.Then total number and total score of bone erosions of each case were calculated according to all affected bones.Taken DECT as reference standard,the efficacy of UTE and T1WI for detecting bone erosions was assessed through comparing with DECT using Kappa coefficient.Results The imaging quality score of T1WI was lower than that of DECT(all P＜0.05),while no significant difference was found between UTE and DECT(all P＞0.05).There was high agreement between UTE and DECT for detecting bone erosions(κ=0.949),while the agreement between T1WI and DECT ranged from good to high(κ=0.718 to 0.805).The total number and total score of bone erosions based on T1WI were significantly lower than those based on DECT(all P＜0.05),while no significant difference was found between UTE and DECT(all P＞0.05).Conclusion UTE was better than T1WI for detecting bone erosions of gouty arthritis.

Objective To investigate the distribution and antimicrobial resistance profiles of common pathogens isolated from cerebrospinal fluid(CSF)in CHINET program from 2015 to 2021.Methods The bacterial strains isolated from CSF were identified in accordance with clinical microbiology practice standards.Antimicrobial susceptibility test was conducted using Kirby-Bauer method and automated systems per the unified CHINET protocol.Results A total of 14 014 bacterial strains were isolated from CSF samples from 2015 to 2021,including the strains isolated from inpatients(95.3％)and from outpatient and emergency care patients(4.7％).Overall,19.6％of the isolates were from children and 80.4％were from adults.Gram-positive and Gram-negative bacteria accounted for 68.0％and 32.0％,respectively.Coagulase negative Staphylococcus accounted for 73.0％of the total Gram-positive bacterial isolates.The prevalence of MRSA was 38.2％in children and 45.6％in adults.The prevalence of MRCNS was 67.6％in adults and 69.5％in children.A small number of vancomycin-resistant Enterococcus faecium(2.2％)and linezolid-resistant Enterococcus faecalis(3.1％)were isolated from adult patients.The resistance rates of Escherichia coli and Klebsiella pneumoniae to ceftriaxone were 52.2％and 76.4％in children,70.5％and 63.5％in adults.The prevalence of carbapenem-resistant E.coli and K.pneumoniae(CRKP)was 1.3％and 47.7％in children,6.4％and 47.9％in adults.The prevalence of carbapenem-resistant Acinetobacter baumannii(CRAB)and Pseudomonas aeruginosa(CRPA)was 74.0％and 37.1％in children,81.7％and 39.9％in adults.Conclusions The data derived from antimicrobial resistance surveillance are crucial for clinicians to make evidence-based decisions regarding antibiotic therapy.Attention should be paid to the Gram-negative bacteria,especially CRKP and CRAB in central nervous system(CNS)infections.Ongoing antimicrobial resistance surveillance is helpful for optimizing antibiotic use in CNS infections.