Objective: To develop a dual-branch deep learning framework for accurate multi-label classification of fundus diseases, addressing the key limitations of insufficient complementary feature extraction and inadequate cross-modal feature fusion in existing automated diagnostic methods. Methods: The fundus multi-label classification dataset with 12 disease categories (FMLC-12) dataset was constructed by integrating complementary samples from Ocular Disease Intelligent Recognition (ODIR) and Retinal Fundus Multi-Disease Image Dataset (RFMiD), yielding 6 936 fundus images across 12 retinal pathology categories, and the framework was validated on both FMLC-12 and ODIR. Inspired by the holistic multi-regional assessment principle of the Five Wheels theory in traditional Chinese medicine (TCM) ophthalmology, the dual-branch multi-label network (DBMNet) was developed as a novel framework integrating complementary visual feature extraction with pathological correlation modeling. The architecture employed a TransNeXt backbone within a dual-branch design: one branch processed red-green-blue (RGB) images to capture color-dependent features, such as vascular patterns and lesion morphology, while the other processed grayscale-converted images to enhance subtle textural details and contrast variations. A feature interaction module (FIM) effectively integrated the multi-scale features from both branches. Comprehensive ablation studies were conducted to evaluate the contributions of the dual-branch architecture and the FIM. The performance of DBMNet was compared against four state-of-the-art methods, including EfficientNet Ensemble, transfer learning-based convolutional neural network (CNN), BFENet, and EyeDeep-Net, using mean average precision (mAP), F1-score, and Cohen's kappa coefficient. Results: The dual-branch architecture improved mAP by 15.44 percentage points over the single-branch TransNeXt baseline, increasing from 34.41% to 44.24%, and the addition of FIM further boosted mAP to 49.85%. On FMLC-12, DBMNet achieved an mAP of 49.85%, a Cohen’s kappa coefficient of 62.14%, and an F1-score of 70.21%. Compared with BFENet (mAP: 45.42%, kappa: 46.64%, F1-score: 71.34%), DBMNet outperformed it by 4.43 percentage points in mAP and 15.50 percentage points in kappa, while BFENet achieved a marginally higher F1-score. On ODIR, DBMNet achieved an F1-score of 85.50%, comparable to state-of-the-art methods. Conclusion DBMNet effectively integrates RGB and grayscale visual modalities through a dual-branch architecture, significantly improving multi-label fundus disease classification. The framework not only addresses the issue of insufficient feature fusion in existing methods but also demonstrates outstanding performance in balancing detection across both common and rare diseases, providing a promising and clinically applicable pathway for standardized, intelligent fundus disease classification.

OBJECTIVE Based on comment texts from the Weibo platform， this study adopts the text mining method to systematically analyze the public’s emotional attitudes and core concerns of “medical insurance reimbursement”， and explore the public’s real experiences and feedback in the implementation of the current system. METHODS With “medical insurance reimbursement” as the keyword， Python web crawler technology was used to collect Weibo comment texts from December 2023 to April 2025. Word frequency analysis， sentiment analysis， semantic network analysis and Latent Dirichlet Allocation （LDA） topic models were adopted to mine the in-depth features of comment texts and investigate the public’s concerns and emotional attitudes toward “medical insurance reimbursement”. RESULTS A total of 3 645 valid comment texts were obtained. The overall emotional of the public toward “medical insurance reimbursement” was weakly positive： 66.4% of the comments contained positive emotions， and 32.4% contained negative emotions. Semantic network analysis indicated that positive public emotions mainly centered on the achievements of medical insurance reform and the improvement of security level， while negative emotions mainly focused on reimbursement rules， economic burden and the implementation of medical insurance policies. The LDA model identified eight core themes， mainly including access of new drugs to medical insurance catalogs， medical insurance support for traditional Chinese medicine and biomedicine， disputes over medical insurance coverage of specific medical items， cross-regional medical treatment and medical insurance settlement， medical insurance reimbursement and medical burden， as well as the approval of innovative drugs and medical insurance payment. CONCLUSIONS The public generally holds a positive attitude toward “medical insurance reimbursement”， yet still has many expectations in terms of policy implementation. On the premise of guaranteeing the sustainability of medical insurance funds， it is suggested to further improve the transparency and fairness of the reimbursement process， and promote the scientific， standardized and efficient development of the medical insurance system. Meanwhile， attention should be paid to public opinion feedback on social media， and the policy communication mechanism should be strengthened， so as to enhance the public’s sense of gain and trust in the medical insurance system.

OBJECTIVE Based on comment texts from the Weibo platform， this study adopts the text mining method to systematically analyze the public’s emotional attitudes and core concerns of “medical insurance reimbursement”， and explore the public’s real experiences and feedback in the implementation of the current system. METHODS With “medical insurance reimbursement” as the keyword， Python web crawler technology was used to collect Weibo comment texts from December 2023 to April 2025. Word frequency analysis， sentiment analysis， semantic network analysis and Latent Dirichlet Allocation （LDA） topic models were adopted to mine the in-depth features of comment texts and investigate the public’s concerns and emotional attitudes toward “medical insurance reimbursement”. RESULTS A total of 3 645 valid comment texts were obtained. The overall emotional of the public toward “medical insurance reimbursement” was weakly positive： 66.4% of the comments contained positive emotions， and 32.4% contained negative emotions. Semantic network analysis indicated that positive public emotions mainly centered on the achievements of medical insurance reform and the improvement of security level， while negative emotions mainly focused on reimbursement rules， economic burden and the implementation of medical insurance policies. The LDA model identified eight core themes， mainly including access of new drugs to medical insurance catalogs， medical insurance support for traditional Chinese medicine and biomedicine， disputes over medical insurance coverage of specific medical items， cross-regional medical treatment and medical insurance settlement， medical insurance reimbursement and medical burden， as well as the approval of innovative drugs and medical insurance payment. CONCLUSIONS The public generally holds a positive attitude toward “medical insurance reimbursement”， yet still has many expectations in terms of policy implementation. On the premise of guaranteeing the sustainability of medical insurance funds， it is suggested to further improve the transparency and fairness of the reimbursement process， and promote the scientific， standardized and efficient development of the medical insurance system. Meanwhile， attention should be paid to public opinion feedback on social media， and the policy communication mechanism should be strengthened， so as to enhance the public’s sense of gain and trust in the medical insurance system.

Objective To develop a rehabilitation exercise program for the patients with osteoporotic vertebral compression fractures.Methods An online search was performed across both Chinese and English databases and websites to retrieve literature on rehabilitation exercises and related therapeutic measures for patients with osteoporotic vertebral compression fractures.Two researchers independently screened the retrieved literature,evaluated the quality,and extracted relevant evidence.Semi-structured interviews were conducted among 15 patients at various rehabilitation phases about the requirements of rehabilitation.A preliminary exercise program was developed and verified for the validity and feasibility using Delphi method.Results A rehabilitation exercise program was formulated.The program included five periods of preoperative period,bedridden period after surgery,ambulation period(1 week after surgery),weeks 2-8 after surgery and weeks 9-12 after surgery covering six domains of exercise type,exercise items,exercise duration,exercise frequency,exercise intensity and exercise safety.The response rate and effective rate over the two rounds of expert consultation were both 100.00%,with an authority coefficient of 0.94.The Kendall's W coefficients for secondary indicators in the two rounds of expert consultation were 0.184 and 0.334,respectively(both P<0.001).Conclusion The exercise program developed in this study for the patients with osteoporotic vertebral compression fractures is scientifically reliable and reasonable.The two features of continuity and staged characteristics in rehabilitation process are fully considered,thereby it offers a guidance for clinical healthcare professionals as well as the patients in development of practical and effective rehabilitation exercise plans.

Objective To develop a rehabilitation exercise program for the patients with osteoporotic vertebral compression fractures.Methods An online search was performed across both Chinese and English databases and websites to retrieve literature on rehabilitation exercises and related therapeutic measures for patients with osteoporotic vertebral compression fractures.Two researchers independently screened the retrieved literature,evaluated the quality,and extracted relevant evidence.Semi-structured interviews were conducted among 15 patients at various rehabilitation phases about the requirements of rehabilitation.A preliminary exercise program was developed and verified for the validity and feasibility using Delphi method.Results A rehabilitation exercise program was formulated.The program included five periods of preoperative period,bedridden period after surgery,ambulation period(1 week after surgery),weeks 2-8 after surgery and weeks 9-12 after surgery covering six domains of exercise type,exercise items,exercise duration,exercise frequency,exercise intensity and exercise safety.The response rate and effective rate over the two rounds of expert consultation were both 100.00%,with an authority coefficient of 0.94.The Kendall's W coefficients for secondary indicators in the two rounds of expert consultation were 0.184 and 0.334,respectively(both P<0.001).Conclusion The exercise program developed in this study for the patients with osteoporotic vertebral compression fractures is scientifically reliable and reasonable.The two features of continuity and staged characteristics in rehabilitation process are fully considered,thereby it offers a guidance for clinical healthcare professionals as well as the patients in development of practical and effective rehabilitation exercise plans.

BackgroundNon-suicidal self-injury （NSSI） behavior among adolescents has become a global public health concern. Anxiety and depression are considered key factors influencing NSSI behavior， while social support may play a protective role in alleviating emotional and behavioral issues. However， existing research has primarily focused on the direct impact of individual factors on NSSI behavior， with insufficient exploration of the combined effects of anxiety， depression and social support. ObjectiveTo investigate the direct effect of anxiety on NSSI， the mediating role of depression and the moderating role of social support in relationship between anxiety and NSSI behavior， thus to provide references for the prevention and intervention of NSSI behavior among adolescents. MethodsIn February 2022， a total of 40 820 students in grades 7 to 12 across 10 middle schools in a district of Chengdu were selected as participants， and they were assessed using Generalized Anxiety Disorder Scale-7 item （GAD-7）， Patient's Health Questionnaire Depression Scale-9 item （PHQ-9）， Social Support Scale for Urban Students （SSSUS） and Adolescent Self-Harm Scale （ASHS）. Pearson correlation analysis was conducted to examine the correlations between scale scores among adolescents with NSSI behaviors. Mediation and moderation analyses were performed using Process 3.5 in SPSS， and the significance was tested with bootstrapping. The interaction was visualized by using simple slope analysis. ResultsAmong 34 534 （84.60%） valid respondents， 542 adolescents （1.57%） reported engaging in NSSI behavior. Significant differences in gender， GAD-7 scores， PHQ-9 scores， and SSSUS scores were observed between NSSI behavior group and non-NSSI group （χ²/t=62.889， 71.120， 94.365， -41.464， P<0.01）.Adolesents with NSSI showed positive correlations between GAD-7 scores and both ASHS and PHQ-9 scores （r=0.158， 0.166， P<0.01）. PHQ-9 scores were positively correlated with ASHS scores （r=0.364， P<0.01）， but negatively correlated with SSSUS scores （r=-0.290， P<0.01）. SSSUS scores were negatively correlated with ASHS scores （r=-0.247， P<0.01）. Depression partially mediated the relationship between anxiety and NSSI behavior， with an effect size of 0.544 （95% CI： 0.162~0.944）， accounting for 35.79% of the total effect. Social support moderated the relationship between depression and NSSI bahavior， with an effect value of -0.082 （95% CI： -0.135~-0.029）. ConclusionAnxiety not only directly influences NSSI bahavior among adolescents， also indirectly exacerbates it through depression， while social support mitigates the impact of depression on NSSI behavior. ［Funded by Youth Project of National Natural Science Foundation of China （number， 82401812）； Project of Health Commission of Sichuan Province （number， 24LCYJPT18）］

With the development of science and technology worldwide, the blooming of artificial intelligence (AI) and big data has brought new opportunities and challenges to the promotion of the research capacity of professional master's students in oncology. The construction of the new medical education in China aims to cultivate high-level medical talents with comprehensive multidisciplinary skills and innovative abilities to flexibly solve complex problems at the frontier of medicine. In this context, professional master's students in oncology, who are facing problems such as low scientific research output and uneven quality and needing improving scientific research literacy, have been required to develop into compound talents with both clinical and research prowess. To cultivate and promote the research capacity of professional master's students in oncology, the key steps include accelerating the construction of AI education and databases, highlighting the cultivation of their scientific research capacity, implementing and fostering the cultivation of innovative ability and scientific research thinking, piloting joint cultivation models by engineering universities and medical universities, emphasizing the construction of the curriculum and teacher team for oncology, piloting the multidisciplinary mode and COME mode, and establishing a multidisciplinary cooperation network.

With the development of science and technology worldwide, the blooming of artificial intelligence (AI) and big data has brought new opportunities and challenges to the promotion of the research capacity of professional master's students in oncology. The construction of the new medical education in China aims to cultivate high-level medical talents with comprehensive multidisciplinary skills and innovative abilities to flexibly solve complex problems at the frontier of medicine. In this context, professional master's students in oncology, who are facing problems such as low scientific research output and uneven quality and needing improving scientific research literacy, have been required to develop into compound talents with both clinical and research prowess. To cultivate and promote the research capacity of professional master's students in oncology, the key steps include accelerating the construction of AI education and databases, highlighting the cultivation of their scientific research capacity, implementing and fostering the cultivation of innovative ability and scientific research thinking, piloting joint cultivation models by engineering universities and medical universities, emphasizing the construction of the curriculum and teacher team for oncology, piloting the multidisciplinary mode and COME mode, and establishing a multidisciplinary cooperation network.

Objective:To examine the factors that contribute to memory impairment in elderly stroke patients and develop a predictive model.Methods:One hundred stroke patients from the First People's Hospital of Changzhou were selected to assess the incidence of memory impairment using the Montreal cognitive assessment memory index score(MoCA-MIS).Univariate analysis and multivariate Logistic regression were performed to determine the factors influencing memory impairment in these patients.Additionally, the correlation among relevant scale scores was examined, and a prediction model was developed.Results:In the study, 49 patients(49.0%)did not exhibit memory impairment.Patients with memory impairment were found to have higher proportions of individuals over 75 years old, elevated levels of triglyceride(TG), total cholesterol(TC), low-density lipoproteins cholesterol(LDL-C), and National Institute of Health Stroke Scale(NIHSS)scores compared to those without memory impairment.Conversely, patients without memory impairment had higher proportions of individuals with more than 9 years of education, higher levels of high-density lipoprotein cholesterol(HDL-C), mini-mental state examination(MMSE)scores, Rivermead behavioural memory test-Ⅱ(RBMT-Ⅱ)scores, and picture-based memory impairment screen(PMIS)scores(all P<0.05).Furthermore, Montreal cognitive assessment-memory index(MoCA-MIS)scores in stroke patients with memory impairment showed negative correlations with NIHSS scores, TG, and LDL-C, while showing positive correlations with HDL-C, MMSE scores, RBMT-Ⅱ scores, and PMIS scores(all P<0.05).Multifactorial Logistic regression analysis indicated that years of education, TG, HDL-C, NIHSS score, MMSE score, RBMT-Ⅱ score, PMIS score, and the location of the lesion in the cortex or temporal lobe were significant factors influencing memory impairment in stroke patients(all P<0.05).The receiver operating characteristic curve(ROC)analysis revealed an area under curve(AUC)of 0.955(95% CI: 0.921-0.977)for the prediction model of memory impairment in stroke patients, with a Yoden index of 0.841. Conclusions:The risk of memory impairment in stroke patients is associated with education years and blood lipid levels.Factors such as high education level, active cognitive function, and memory training serve as protective factors against memory impairment.The prediction model developed using these influencing factors demonstrates high predictive accuracy for post-stroke memory impairment.