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 To explore the relationship between ambulatory blood pressure changes and health-related behaviors as well as sleep quality in hypertensive patients. Methods A retrospective study was conducted on 684 patients who underwent 24-hour ambulatory blood pressure monitoring at Chengdu Fifth People's Hospital between January 2023 and April 2025. The cohort included 502 hypertensive patients and 182 non-hypertensive patients. Clinical data from both groups of patients were collected. Their health-related behaviors and sleep quality were investigated. Pearson correlation analysis was conducted to explore the relationship between ambulatory blood pressure changes and health-related behaviors as well as sleep quality in hypertensive patients. Results The differences in dSBP, dDBP, nSBP, nDBP, 24-hour SBP, 24-hour DBP, 24-hour SBP CV, and 24-hour DBP CV levels between the two groups were statistically significant (P<0.05). The scores of the Active Health Behavior Scale (AHBS) in the hypertension group were lower than those in the non-hypertension group, while the scores of the Pittsburgh Sleep Quality Index (PSQI) in the hypertension group were higher than those in the non-hypertension group (P<0.05). AHBS showed a negative correlation with dSBP, dDBP, nSBP, nDBP, 24 h SBP, 24 h DBP, 24 h SBP CV, and 24 h DBP CV (P<0.05), while PSQI showed a positive correlation with dSBP, dDBP, nSBP, nDBP, 24 h SBP, 24 h DBP, 24 h SBP CV, and 24 h DBP CV (P<0.05). Conclusion Health-related behaviors and sleep quality can influence ambulatory blood pressure through various mechanisms, including neural, endocrine, and vascular functions.

ObjectiveTo clarify whether METTL14 mediates the core role of acupuncture at Neiguan (PC6) in promoting myelination and improving behavior in young autistic rats through gene intervention technology. MethodsThe ASD model was established by intraperitoneal injection of valproic acid (VPA) in pregnant rats. Male offspring were intracerebroventricularly injected with adenovirus-packaged METTL14 shRNA (sh-METTL14) or its control (sh-NC) on postnatal day 1, with a model group set as well. Subsequently, the juvenile rats were divided into model group, acupuncture group, acupuncture+sh-NC group, and acupuncture+sh-METTL14 group. The acupuncture group received acupuncture at Neiguan (PC6) from postnatal day 7, once daily for 21 consecutive days. Neurobehavioral changes were evaluated by behavioral tests; METTL14 knockdown efficiency and the expression of METTL14, METTL3, and PTEN were detected by quantitative real-time PCR (qRT-PCR) and Western blot (WB); PTEN m⁶A levels were measured by RNA immunoprecipitation-qPCR (RIP-qPCR); myelin ultrastructure, expression of myelin basic protein (MBP) and neurofascin 155 (NF155), and dendritic spine density were observed using transmission electron microscopy (TEM), enzyme-linked immunosorbent assay (ELISA), immunofluorescence, qRT-PCR, and primary neuron culture. ResultsBehaviorally, knockdown of METTL14 significantly counteracted the beneficial effects of acupuncture in improving self-grooming, open field exploration, three-chamber social interaction, and Morris water maze learning and memory (P<0.05, P<0.01). Compared with the acupuncture+sh-NC group, the acupuncture+sh-METTL14 group showed significantly decreased mRNA and protein expression of hippocampal METTL14 (P<0.01), and the upregulating effects of acupuncture on METTL3 and PTEN expression were reversed (P<0.01). Meanwhile, knockdown of METTL14 significantly inhibited the acupuncture-induced increase in PTEN m⁶A levels (P<0.01). Morphologically, knockdown of METTL14 attenuated the improvement of myelin structure by acupuncture, reversed the downregulation of MBP and upregulation of NF155 induced by acupuncture, and blocked the increase in dendritic spine density (P<0.05, P<0.01). ConclusionMETTL14 is a key molecule mediating the therapeutic effect of acupuncture at Neiguan. Acupuncture at Neiguan upregulates METTL14, thereby enhancing m⁶A methylation modification of PTEN mRNA to stabilize its expression, ultimately promoting myelin development and improving behavioral symptoms in ASD juvenile rats. This preliminarily reveals the modern biological connotation of “opening Xuanfu and dredging myelin”.

ObjectiveTo clarify whether METTL14 mediates the core role of acupuncture at Neiguan (PC6) in promoting myelination and improving behavior in young autistic rats through gene intervention technology. MethodsThe ASD model was established by intraperitoneal injection of valproic acid (VPA) in pregnant rats. Male offspring were intracerebroventricularly injected with adenovirus-packaged METTL14 shRNA (sh-METTL14) or its control (sh-NC) on postnatal day 1, with a model group set as well. Subsequently, the juvenile rats were divided into model group, acupuncture group, acupuncture+sh-NC group, and acupuncture+sh-METTL14 group. The acupuncture group received acupuncture at Neiguan (PC6) from postnatal day 7, once daily for 21 consecutive days. Neurobehavioral changes were evaluated by behavioral tests; METTL14 knockdown efficiency and the expression of METTL14, METTL3, and PTEN were detected by quantitative real-time PCR (qRT-PCR) and Western blot (WB); PTEN m⁶A levels were measured by RNA immunoprecipitation-qPCR (RIP-qPCR); myelin ultrastructure, expression of myelin basic protein (MBP) and neurofascin 155 (NF155), and dendritic spine density were observed using transmission electron microscopy (TEM), enzyme-linked immunosorbent assay (ELISA), immunofluorescence, qRT-PCR, and primary neuron culture. ResultsBehaviorally, knockdown of METTL14 significantly counteracted the beneficial effects of acupuncture in improving self-grooming, open field exploration, three-chamber social interaction, and Morris water maze learning and memory (P<0.05, P<0.01). Compared with the acupuncture+sh-NC group, the acupuncture+sh-METTL14 group showed significantly decreased mRNA and protein expression of hippocampal METTL14 (P<0.01), and the upregulating effects of acupuncture on METTL3 and PTEN expression were reversed (P<0.01). Meanwhile, knockdown of METTL14 significantly inhibited the acupuncture-induced increase in PTEN m⁶A levels (P<0.01). Morphologically, knockdown of METTL14 attenuated the improvement of myelin structure by acupuncture, reversed the downregulation of MBP and upregulation of NF155 induced by acupuncture, and blocked the increase in dendritic spine density (P<0.05, P<0.01). ConclusionMETTL14 is a key molecule mediating the therapeutic effect of acupuncture at Neiguan. Acupuncture at Neiguan upregulates METTL14, thereby enhancing m⁶A methylation modification of PTEN mRNA to stabilize its expression, ultimately promoting myelin development and improving behavioral symptoms in ASD juvenile rats. This preliminarily reveals the modern biological connotation of “opening Xuanfu and dredging myelin”.

ObjectiveAs a common lifestyle habit, alcohol consumption has a controversial association with the onset of Parkinson’s disease (PD). To demonstrate the correlation between alcohol consumption and PD and to identify associated genes, we integrated findings from clinical surveys, genomics, transcriptomics, and animal experiments. MethodsWe investigated the alcohol consumption rates (including both before and after disease onset) among 244 PD patients in China and 177 PD patients from the U.S. NHANES database. Mendelian randomization (MR) analysis was performed using genome-wide association study (GWAS) data for three alcohol-related traits and seven PD-related datasets from the MRC IEU OpenGWAS database. Transcriptomic data from the substantia nigra of PD patients were obtained from three GEO datasets (GSE7621, GSE20141, and GSE49036) to analyze RIT2 gene transcription. Finally, three groups of animal experiments (water/20% ethanol/20% liquor, with 4 C57BL/6J mice per group) were conducted to examine changes in brain RIT2 gene expression and transcriptomic profiles following alcohol consumption. ResultsThe alcohol consumption rates among PD patients in China and the U.S. (9%-18.87%) were significantly lower than the general population rates of 15%-45% in their respective regions (P<0.001), suggesting a possible negative association between alcohol consumption and PD. Subsequently, in 21 bidirectional MR analyses using 3 alcohol-related GWAS datasets and 7 PD-related GWAS datasets, the forward MR analyses (alcohol intake as exposure, PD as outcome) yielded 12 negative associations (OR_IVW<1) and 9 positive associations (OR_IVW>1). Among these, only two negative associations reached statistical significance: alcohol intake frequency (OR_IVW=0.75, 95% CI: 0.60-0.93, P=0.010) and alcohol consumption (OR_IVW=0.20, 95% CI: 0.05-0.83, P=0.026). The forward MR analysis (alcohol intake→PD) identified 235 SNPs, annotated to 316 genes, while the reverse MR analyses (PD→alcohol intake) identified 37 SNPs, annotated to 53 genes. Notably, only the RIT2 gene appeared in both the forward and reverse MR analyses (alcohol intake→PD: rs28597806, rs8083110; PD→alcohol intake: rs4588066). RIT2 is selectively expressed in the human brain (FPKM: 5.259±2.103), with low or no expression in peripheral tissues (FPKM: <1). Analysis of three human substantia nigra transcriptomic datasets revealed a decreasing trend in RIT2 gene expression in PD patients (GSE20141 array signal: 3.49±1.23 vs. 2.33±0.87, P=0.044). Animal experiments demonstrated that administration of 20% ethanol or 20% liquor (approximately 8% ethanol) stimulated a >2-fold upregulation of RIT2 gene expression in the mouse brain. Furthermore, transcriptomic sequencing revealed that the two alcohol-treated groups exhibited 96 (20% ethanol vs. water control) and 4 (20% liquor vs. water control) differentially expressed genes, respectively, indicating that low-dose alcohol consumption can achieve RIT2 upregulation while minimizing impact on other brain genes. In addition to its anti-infective effects, low-dose alcohol consumption primarily influences signaling pathways related to neurodegenerative diseases such as PD and Prion diseases. ConclusionAlcohol consumption is generally considered as a harmful lifestyle habit. However, some studies have also shown a lower risk of mortality among individuals who consume low doses of alcohol (100 g/week of ethanol) or drink occasionally. Currently, one of the research focuses on alcohol consumption is whether the human body can benefit from low-dose alcohol intake. This study provides new evidence supporting a negative association between alcohol consumption and PD, and for the first time, through MR analysis, identifies the RIT2 gene as a potential mediator of the effect of alcohol consumption on PD. RIT2 is selectively expressed in the human brain. Building upon existing evidence indicating downregulated RIT2 gene expression in PD pathogenesis, our experiments confirm that low-dose alcohol consumption can upregulate RIT2 expression in the brain. In brief, alcohol consumption may suppress the pathogenesis of PD by upregulating RIT2 expression in the substantia nigra. China is facing a serious problem of population aging. This study offers important insights for long-term PD prevention and treatment strategies, with the aim of benefiting more potential PD patients through lifestyle modifications, thereby improving the quality of life of the aging population and reducing the economic burden on healthcare.

Objective To analyze the infection status and epidemic trends of soil-borne nematode infections in Yubei, Chongqing City, in 2010,2021, and 2022. Methods The local populations from four survey sites of four towns in 2010 and five sites of five towns in 2021 and 2022 were surveyed regarding their basic information using a unified form. Fecal samples of the participants were collected and tested for soil-borne nematode infections using the modified Kato-Katz thick smear method. Results In 2010, 2 049 participants were surveyed, followed by 1 000 participants in 2021 and 2022. The overall prevalence of parasitic infections declined significantly from 3.86% to 0.20%. In 2010, soil-transmitted nematode included hookworms(3.81%) and roundworms(0.29%). In 2021, the infection rates of roundworms and hookworms were 1.70% and 0.10% respectively. Notably, only Ascaris was identified in 2022(0.20%). The≥60 age group consistently exhibited the highest infection rates across all surveys, followed by the 40-59 age group. The infection rates of males in the three surveys were 3.31%,1.92%, and 0.20% respectively, and those of females were 4.37%, 1.46%, and 0.20% respectively. There was no statistically significant difference in the infection rates between males and females. Educational attainment was inversely associated with infection; in 2010, the highest prevalence was observed among those with primary education or below, whereas in 2021, illiterate or semi-literate individuals showed the highest susceptibility. The occupational distribution of infections in 2010 indicated that retirees (8.33%), farmers(4.86%), and homemakers or unemployed individuals(3.45%)were the most affected. However, in 2021 and 2022, farmers emerged as the predominant occupational group with soil-transmitted nematode infections. Conclusions The infection rate of soil-borne nematodes showed a decreasing trend in Yubei, and the infection species changed from hookworms in 2010 to Ascaris in 2022. Farmers, the elderly, and people with low education levels should continue to be the focus of preventive and control efforts.

Objective To investigate the status and predictive factors of future acute exacerbation of asthma in adults. Methods Adult patients with asthma who were admitted to the hospital from March 2023 to July 2025 were selected. Based on the acute exacerbation status after follow-up, the patients were categorized into acute exacerbation group (n=107) and control group (n=245). The clinical data, pulmonary function [peak expiratory flow (PEF) and forced expiratory volume in the 1st second/forced vital capacity (FEV₁/FVC)] and health status [asthma control questionnaire (ACQ), Morisky medication adherence scale (MMAS-8), short form 36-item health survey (SF-36)] of the study subjects were collected. The correlation between the indicators and acute exacerbation was explored by logistic regression analysis, and the predictive efficiency was evaluated by ROC curve. Results The age, smoking history, asthma course, allergen exposure and ACQ in the acute exacerbation group were higher than those in the control group, while the average monthly family income, glucocorticoid treatment, PEF, FEV₁/FVC, MMAS-8 and SF-36 were lower (P<0.05). Multivariate logistic regression suggested that ACQ score, glucocorticoid treatment, MMAS-8 score, FEV₁/FVC, PEF and smoking history were the influencing factors of acute asthma exacerbation (P<0.05). ROC analysis revealed that the diagnostic efficiency of combined prediction was the highest (AUC=0.965), with sensitivity of 0.944 and specificity of 0.857 (P<0.05). Conclusion It is necessary to pay more attention to the medication compliance and asthma control level of patients, and integrate multi-dimensional information such as smoking history, pulmonary function and treatment conditions to accurately identify patients with high-risk acute exacerbation and implement targeted intervention measures.

ObjectivePeripheral nerve injury is a common clinical condition that can impair the regulatory function of target organs, leading to functional disability; severe injuries may even result in lifelong disability. Therefore, how to effectively promote the morphological and functional reconstruction of injured peripheral nerves has become a critical scientific issue that urgently needs to be addressed. This study biomimicked the highly oriented structure and electrophysiological properties of peripheral nerves and investigated the feasibility and effectiveness of combining electrical stimulation (ES) with oriented conductive substrates for the repair of injured peripheral nerves at the cellular level. MethodsPolydimethylsiloxane (PDMS) substrates with oriented microgroove structures of varying widths (0, 5, 10, and 20 μm) were fabricated using silicon wafers with micropatterned grooves as templates. Graphene-based conductive patterned substrates (GCPSs) with oriented microgrooves were then constructed via soft lithography and designated as GCPS 0, GCPS 5, GCPS 10, and GCPS 20 according to the groove width. The GCPSs were prepared from collagen/polycaprolactone (PCL) matrices containing 1.0 wt% graphene, and their morphology, wettability, and conductivity were characterized. Mesenchymal stem cells (MSCs) viability, proliferation, adhesion, and orientation on the substrates were evaluated before ES treatment. Then, ES at a frequency of 2 Hz and field strengths of 10, 20, or 50 mV/cm were applied to the cells for 10 min daily, over 3 or 7 consecutive days. Neural differentiation was assessed by morphological observation, immunofluorescence staining, and analysis of neural lineage-related markers. ResultsThe oriented microgroove PDMS substrates supported long-term MSCs growth, guided cells to align along the groove direction, and promoted a morphological transition toward neural-like cells, exhibiting the capacity to induce MSCs differentiation toward the neural lineage. Microgroove structures effectively guided the alignment of MSCs and promoted the development of a more elongated, neural-like morphology, with the most pronounced morphological changes observed on grooves of 10 and 20 μm in width. Meanwhile, the incorporation of graphene endowed the GCPSs with favorable electrical conductivity, though the groove width did not significantly affect the conductive performance. The GCPSs showed good cytocompatibility with MSCs, and cell proliferation on these substrates exhibited a time-dependent pattern, with no significant effect observed from groove width. The inductive effect of ES combined with GCPSs on MSCs differentiation toward specific neural lineages was co-regulated by ES intensity (0, 10, 20, and 50 mV/cm) and microgroove width. ES at 10 mV/cm and 20 mV/cm promotes MSCs differentiation toward neurons on microgrooved substrates of various widths (5, 10, and 20 μm); under the same ES intensities, microgrooves with widths of 5 μm and 10 μm are more favorable for MSCs differentiation toward glial cells/SCs. Notably, low-intensity ES (10 or 20 mV/cm) combined with GCPS 10 significantly enhanced the neural differentiation of MSCs. Neurotrophic factor secretion levels by MSCs cultured on various GCPS substrates after 3 d of ES application was further detected. There were no significant differences in nerve growth factor‑β (NGF‑β) and brain-derived neurotrophic factor (BDNF) concentrations among the GCPS 5, GCPS 10, and GCPS 20 substrates, and these levels were not affected by variations in ES intensity. Under ES at 10 mV/cm, the glial cell line-derived neurotrophic factor (GDNF) concentration in each GCPS substrate was higher than that in the corresponding non-ES group on the same substrate. However, as ES intensity increased, the GDNF concentration decreased to varying degrees. ConclusionThis study demonstrates that the combined strategy of “oriented topographical structure+ES” can effectively promote the neural differentiation of MSCs and holds promise for accelerating the repair process of peripheral nerve injuries.

ObjectiveThis study examined KCCQ symptom score trajectories during the vulnerable phase of heart failure and their association with readmission within 6 months. MethodsThis prospective cohort study included 1 109 patients hospitalized for chronic heart failure at a tertiary hospital in Nanchang， China， between August 2020 and July 2025. KCCQ scores were assessed at six time points： day 3 after admission， discharge， and 1， 2， 3， and 6 months after discharge. Group-based trajectory modeling was used to identify patterns of change in KCCQ scores. Multiple logistic regression was used to assess the association between trajectory membership and readmission risk. ResultsDuring follow-up， 229 patients were readmitted， accounting for 20.6% of the cohort. Four KCCQ trajectories were identified： persistently declining scores （n=228， 20.6%）， persistently improving scores （n=110， 9.9%）， stable intermediate scores （n=545， 49.1%）， and stable high scores （n=226， 20.4%）. Compared with the stable intermediate group， the persistently declining group had a higher risk of readmission in the fully adjusted model （OR=2.099， 95% CI： 1.027–4.315， P=0.043）， whereas the stable high-score group had a lower risk （OR=0.260， 95% CI： 0.127–0.494， P<0.001）. The association between the persistently improving trajectory and readmission was not statistically significant （OR=0.614， 95% CI： 0.306–1.154， P=0.147）. ConclusionsKCCQ symptom score trajectories during the vulnerable phase of heart failure show marked heterogeneity and are associated with readmission within 6 months. A persistently declining trajectory is associated with higher readmission risk， whereas a stable high-score trajectory is associated with lower risk. Longitudinal monitoring of KCCQ score changes may help inform post-discharge risk stratification and follow-up management in patients with heart failure.

ObjectiveThis study examined KCCQ symptom score trajectories during the vulnerable phase of heart failure and their association with readmission within 6 months. MethodsThis prospective cohort study included 1 109 patients hospitalized for chronic heart failure at a tertiary hospital in Nanchang， China， between August 2020 and July 2025. KCCQ scores were assessed at six time points： day 3 after admission， discharge， and 1， 2， 3， and 6 months after discharge. Group-based trajectory modeling was used to identify patterns of change in KCCQ scores. Multiple logistic regression was used to assess the association between trajectory membership and readmission risk. ResultsDuring follow-up， 229 patients were readmitted， accounting for 20.6% of the cohort. Four KCCQ trajectories were identified： persistently declining scores （n=228， 20.6%）， persistently improving scores （n=110， 9.9%）， stable intermediate scores （n=545， 49.1%）， and stable high scores （n=226， 20.4%）. Compared with the stable intermediate group， the persistently declining group had a higher risk of readmission in the fully adjusted model （OR=2.099， 95% CI： 1.027–4.315， P=0.043）， whereas the stable high-score group had a lower risk （OR=0.260， 95% CI： 0.127–0.494， P<0.001）. The association between the persistently improving trajectory and readmission was not statistically significant （OR=0.614， 95% CI： 0.306–1.154， P=0.147）. ConclusionsKCCQ symptom score trajectories during the vulnerable phase of heart failure show marked heterogeneity and are associated with readmission within 6 months. A persistently declining trajectory is associated with higher readmission risk， whereas a stable high-score trajectory is associated with lower risk. Longitudinal monitoring of KCCQ score changes may help inform post-discharge risk stratification and follow-up management in patients with heart failure.