ObjectiveCerebrospinal fluid (CSF) plays a crucial role in maintaining the homeostasis of the central nervous system (CNS). CSF rapidly exchanges with interstitial fluid (ISF) via the glymphatic system within the brain parenchyma. CSF-ISF circulation and its associated mechanisms are often referred to as the brain lymphatic system. This system is connected directly to meningeal lymphatic vessels (mLVs), jointly performing the function of clearing metabolic waste from the CNS. Emerging evidence indicates that this system is closely associated with the onset and progression of neurodegenerative diseases (NDs) such as Alzheimer’s disease (AD). Importantly, abnormal CSF circulation is not only a downstream consequence of AD pathology, but also a risk factor. In AD, the dynamics of CSF flow within the CNS are diminished, immune dysregulation occurs, and this may increase the risk of AD by exacerbating the burden of amyloid β-protein (Aβ). In the mouse model of AD, impaired CSF flow compromises this clearance function, leading to cognitive deficits. Clinically, acupuncture at cognition-related acupoints is commonly used for the prevention and treatment of AD. However, whether its therapeutic effects are mediated through the modulation of CSF dynamics remains unclear. This study aimed to evaluate the impact of acupuncture on CSF flow and investigate its acupoint specificity. MethodsMice were randomly assigned to experimental groups for the different electroacupuncture groups with the following acupoints: Baihui point (GV 20), Ear point, Neiguan point (PC 6), and Tianshu point (ST 25). Wild-type mice on a C57BL/6J background were used as controls. Fluorescent tracer was injected into the cisterna magna to label CSF flow. Fluorescence imaging was employed to assess the distribution of CSF within the brain before and after acupuncture stimulation. ResultsFollowing tracer injection into the cisterna magna, fluorescence signals rapidly reached the cerebellum and medulla—the regions closest to the injection site. Fluorescence intensity was higher in ventral brain regions compared to dorsal regions, likely due to greater vascular density in ventral areas facilitating CSF-ISF exchange. Electroacupuncture at the GV 20 produced the most pronounced enhancement of CSF across the whole brain, while stimulation at the ST 25 primarily augmented flow within subcortical regions. In contrast, electroacupuncture at the Ear point or the PC 6 had no observable effect on CSF in mice. ConclusionElectroacupuncture promotes CSF flow into the brain parenchyma in an acupoint-specific manner, with GV 20 exhibiting the most pronounced enhancement of CSF dynamics. These findings suggest that acupuncture-mediated facilitation of CSF flow may represent a potential therapeutic strategy for preventing or delaying age-related cognitive decline.

ObjectiveCerebrospinal fluid (CSF) plays a crucial role in maintaining the homeostasis of the central nervous system (CNS). CSF rapidly exchanges with interstitial fluid (ISF) via the glymphatic system within the brain parenchyma. CSF-ISF circulation and its associated mechanisms are often referred to as the brain lymphatic system. This system is connected directly to meningeal lymphatic vessels (mLVs), jointly performing the function of clearing metabolic waste from the CNS. Emerging evidence indicates that this system is closely associated with the onset and progression of neurodegenerative diseases (NDs) such as Alzheimer’s disease (AD). Importantly, abnormal CSF circulation is not only a downstream consequence of AD pathology, but also a risk factor. In AD, the dynamics of CSF flow within the CNS are diminished, immune dysregulation occurs, and this may increase the risk of AD by exacerbating the burden of amyloid β-protein (Aβ). In the mouse model of AD, impaired CSF flow compromises this clearance function, leading to cognitive deficits. Clinically, acupuncture at cognition-related acupoints is commonly used for the prevention and treatment of AD. However, whether its therapeutic effects are mediated through the modulation of CSF dynamics remains unclear. This study aimed to evaluate the impact of acupuncture on CSF flow and investigate its acupoint specificity. MethodsMice were randomly assigned to experimental groups for the different electroacupuncture groups with the following acupoints: Baihui point (GV 20), Ear point, Neiguan point (PC 6), and Tianshu point (ST 25). Wild-type mice on a C57BL/6J background were used as controls. Fluorescent tracer was injected into the cisterna magna to label CSF flow. Fluorescence imaging was employed to assess the distribution of CSF within the brain before and after acupuncture stimulation. ResultsFollowing tracer injection into the cisterna magna, fluorescence signals rapidly reached the cerebellum and medulla—the regions closest to the injection site. Fluorescence intensity was higher in ventral brain regions compared to dorsal regions, likely due to greater vascular density in ventral areas facilitating CSF-ISF exchange. Electroacupuncture at the GV 20 produced the most pronounced enhancement of CSF across the whole brain, while stimulation at the ST 25 primarily augmented flow within subcortical regions. In contrast, electroacupuncture at the Ear point or the PC 6 had no observable effect on CSF in mice. ConclusionElectroacupuncture promotes CSF flow into the brain parenchyma in an acupoint-specific manner, with GV 20 exhibiting the most pronounced enhancement of CSF dynamics. These findings suggest that acupuncture-mediated facilitation of CSF flow may represent a potential therapeutic strategy for preventing or delaying age-related cognitive decline.

Objective To assess the survival vulnerability of Oncomelania hupensis in Jiangxi Province under future climate scenarios, and to identify low-vulnerability areas for its survival in this province. Methods Village-level O. hupensis snail survey and O. hupensis snail control with chemical treatments in Jiangxi Province from 2016 to 2024 were captured from the Parasitic Disease Prevention and Control Information Management System of China Disease Prevention and Control Information System. Climatic data were primarily sourced from the Resource and Environmental Science Data Platform, Chinese Academy of Sciences (http://www.resdc.cn/), including annual average temperature, annual average precipitation, annual accumulated temperature above 10 °C, annual accumulated temperature above 0 °C, annual maximum temperature, annual minimum temperature, and annual average relative humidity, and nineteen bioclimatic variables were downloaded from the WorldClim website (https://www.worldclim.org/), including mean diurnal range, isothermality, temperature seasonality, and so on. Elevation and normalized difference vegetation index were catprued from the Resource and Environmental Science Data Platform, Chinese Academy of Sciences (http://www.resdc.cn/), and distance to rivers was downloaded from the WorldPop website (http://www.worldpop.org), and land use and land cover (LULC) data were downloaded from the Big Earth Data Center, Chinese Academy of Sciences (https://data.casearth.cn/), and nature reserve data were obtained from the China Nature Reserve Specimen Resource Sharing Platform (http://www.papc.cn/). Three Shared Socioeconomic Pathways (SSPs) from the Beijing Climate Center-Climate System Model version 2-Medium Resolution (BCC-CSM2-MR) global climate model were employed as future climate scenarios, including SSP126, SSP245, SSP585, and the biomod2 ensemble model in R package was used to simulate suitable habitats for O. hupensis snails in Jiangxi Province in 2050 and 2070 under these scenarios. A snail survival vulnerability index was constructed based on the area of suitable snail habitats, area covered by snail control through chemical treatment, area covered by nature reserves, and changes in snail habitat fragmentation, and a map of snail survival vulnerability distribution was plotted. Results The real area of snail habitats ranged from 78 486.76 to 85 309.47 hm², and the area of snail control with chemical treatment ranged from 10 138.98 to 13 240.16 hm² in Jiangxi Province from 2016 to 2024. There were 429 to 531 villages detected with snails during the nine-year period, and the number of actually snail-infested villages ranged from 645 to 686. A total of 818 snail-present points and 1 996 snail-absent points were obtained from snail survey records. The best performance of the biomod2 ensemble model was achieved if a weighted mean approach was used as the ensemble strategy, with a true skill statistic value of 0.799 and an area under the receiver operating characteristic curve of 0.957, and modeling identified annual average relative humidity and annual average precipitation as two most influencing climatic variables for snail distribution. Relative to the current areas of suitable snail habitats under present climate conditions, the area of suitable snail habitats was projected to expand by 24.49% to 46.28% in Jiangxi Province under future climate scenarios, and the proportion of nature reserves areas in the areas of suitable snail habitats was projected to decrease slightly from the current 2.77% to approximately 2.52%, while the proportion of areas of snail control through chemical treatment in areas of suitable snail habitats varied from 0.64% to 19.57%, and the percentage of changes in snail habitat fragmentation ranged from 3.86% to 12.23%. Based on these four indicators, the snail survival vulnerability index was estimated to range from –1.96 to 0.62 in Jiangxi Province. The arithmetic mean of the snail survival vulnerability index differed under three SSP scenarios (SSP126, SSP245 and SSP585), with the highest mean value (–0.69) in 2070 under SSP126, and the lowest mean value (–0.78) in 2070 under SSP585. Conclusions The snail survival vulnerability index ranges from –1.96 to 0.62 in Jiangxi Province under future climate scenarios, and the suitable habitats for O. hupensis snails appear an overall tendency towards expansion. Low-vulnerability snail habitats are mainly distributed along the shores of Poyang Lake and the Yangtze River in Jiangxi Province, partially overlapping with nature reserves. Intensified surveillance of O. hupensis snails is recommended in these areas in the future.

With the accelerating trend of population aging, the number of elderly patients with lung cancer continues to rise, and the disease burden is becoming increasingly heavy. The clinical management of these patients faces severe challenges due to their decreased physiological reserve, complex comorbidities, and significant individual heterogeneity. Consequently, under traditional diagnosis and treatment models, doctors often struggle to identify the individualized risks of elderly patients in a timely and comprehensive manner, which can easily lead to decision biases such as undertreatment or overtreatment. In view of this, this study advocates for the establishment of an umbrella decision-making model specifically tailored for elderly lung cancer patients. Grounded in a multidisciplinary team (MDT) platform, this model deeply integrates oncological indicators with the comprehensive geriatric assessment (CGA) system. By holistically considering multidimensional variables including tumor burden, organ function, frailty index, cognitive status, and social support, the model establishes an operational mechanism characterized by "single entry, precise stratification, and targeted selection". Accordingly, patients can be scientifically triaged into distinct intervention tiers, such as active surveillance, minimally invasive surgery, drug therapy, radiotherapy, and best supportive care, thereby achieving real-time alignment between treatment intensity and patient fitness. This article elaborates on the construction logic and key operational procedures of this novel decision-making framework, aiming to guide clinical practice beyond the limitations of a tumor-centric perspective toward a holistic, dynamic, whole-course management strategy. This transition seeks to ensure optimal quality of life and clinical net benefit for elderly patients alongside survival prolongation.

ObjectiveTo observe the clinical efficacy of electroacupuncture in treatment of simple obesity patients and its effects on the metabolism-inflammation network. MethodsSeventy-two patients with simple obesity were randomly divided into a electroacupuncture group and a sham electroacupuncture group， with 36 patients in each group. In the electroacupuncture group， conventional acupuncture was applied at Zhongwan （CV 12）， bilateral Tianshu （ST 25）， bilateral Daheng （SP15）， Qihai （CV 6）， Guanyuan （CV 4）， bilateral Zusanli （ST 36） and Fenglong （ST 40）. The sham electroacupuncture group used placebo needle at same acupoints as the electroacupuncture group. Both groups were given sparse-dense waves with a frequency of 2/10 Hz， 30 minutes per session， 3 times a week， for a total of 8 weeks of treatment. The changes of human body parameters including body weight， body mass index （BMI）， waist circumference， and body fat percentage （BFP）， metabolic factors including serum fasting blood glucose （FBG）， triglyceride （TG）， total cholesterol （TC）， high density lipoprotein （HDL）， low density lipoprotein （LDL）， leptin， and ghrelin， and inflammatory factors including interleukin-6 （IL-6） and tumor necrosis factor α （TNF-α） were compared before and after treatment. Correlations between changes in leptin， ghrelin and inflammatory factors were further analyzed. ResultsWaist circumference was significantly reduced after treatment in both groups compared with baseline （P＜0.05）， and the post-treatment waist circumference was lower in the electroacupuncture group than in the sham electroacupuncture group （P＜0.05）； body weight， BMI and BFP decreased significantly after treatment compared to pretreatment levels in the electroacupuncture group （P＜0.05）. Serum leptin， IL-6 and TNF-α declined significantly after treatment in both groups versus baseline （P＜0.05）； the electroacupuncture group had lower post-treatment levels of leptin， IL-6 and TNF-α as well as larger pre-post changes than the sham group （P＜0.05）； ghrelin increased markedly after treatment in both groups （P＜0.05）， with higher post-treatment ghrelin and greater pre-post variation observed in the electroacupuncture group （P＜0.05）. In the electroacupuncture group， changes in ghrelin were negatively correlated with IL-6 and positively correlated with TNF-α （P＜0.01）； in the sham electroacupuncture group， ghrelin changes were negatively correlated with both IL-6 and TNF-α， whereas leptin changes were positively correlated with the two inflammatory markers （P＜0.01）. ConclusionElectroacupuncture can effectively improve clinical manifestations， as well as regulate serum metabolic and pro-inflammatory factors in patients with simple obesity， and discrepant correlation patterns of metabolism-inflammation are observed between the two groups.

ObjectiveThe frequency difference limen (FDL) serves as a fundamental metric utilized for effectively quantifying the precise perceptual capabilities of the central auditory system. However, traditional measurement methods rely heavily on the active behavioral responses of subjects and are consequently highly susceptible to the negative influence of confounding subjective factors. Furthermore, existing research paradigms frequently employ uniform stimulus configurations that overlook critical individual perceptual differences. Based on brain-computer interface (BCI) technology, this comprehensive study aims to establish an objective and quantitative evaluation method for auditory frequency discrimination by systematically analyzing and decoding the specific neural responses elicited at the exact threshold state. MethodsWe designed a personalized rapid serial auditory presentation (RSAP) paradigm customized based on each individual’s precise FDL. A cohort of eleven healthy participants was recruited to evaluate the paradigm using pure-tone sequences at a baseline frequency of 4 000 Hz. This experimental paradigm simulates a realistic auditory perception environment through the continuous presentation of acoustic stimuli, thereby allowing for an in-depth investigation into the specific neural representations evoked by weak frequency deviations at the threshold state. Given that auditory stimulus-evoked response features exhibit complex and differentiated spatiotemporal distribution patterns across multiple frequency domains, this study further deeply integrates the cross-scale feature interaction module with the dynamic spatiotemporal attention allocation strategy, innovatively proposing the Multi-Scale Spatial-Temporal Dual Attention Network (MS-STAMNet). Specifically, the network constructs parallel processing branches with multiple receptive fields and introduces a dynamic adaptive weighting strategy to precisely localize core neural activity signals, further deeply integrating multi-scale information through cross-branch feature information interaction to achieve robust single-trial decoding of weak auditory evoked responses. ResultsThe comprehensive electrophysiological data analysis demonstrated that subtle auditory frequency deviation stimuli presented at the threshold level successfully elicited pronounced N2 and P3 event-related potential features, reflecting pre-attentive mismatch detection and subsequent cognitive evaluation, which were prominently distributed over the frontal, central, and temporal regions of the scalp. In the complex time-frequency domain, the extracted neural response characteristics exhibited distinct, statistically significant event-related synchronization within both the low-frequency δ and θ frequency bands, which was simultaneously accompanied by a widespread, prominent event-related desynchronization within the higher α band. A comparative analysis of model performance demonstrated that MS-STAMNet achieved an average unweighted average recall (UAR) of (69.67±6.12)% and area under the curve (AUC) of 0.761 8±0.07, significantly outperforming the established baseline models such as EEGNet and PLNet. Furthermore, a distinct dissociation phenomenon was verified between neural decoding and behavioral performance through regression analysis (R²=0.016, P=0.709), indicating that this model can effectively capture the implicit features of subtle frequency deviations, even when they fail to trigger explicit conscious responses. Additionally, attention weight visualization analysis further reveals the highly accurate focus of the network on key features concentrated over the bilateral temporal and fronto-parietal regions. ConclusionThis study systematically and comprehensively uncovers the multi-dimensional spatiotemporal evolutionary patterns of complex neural responses processing subtle acoustic variations under long-sequence threshold auditory stimulation. Concurrently, it verifies the efficacy and robustness of the proposed MS-STAMNet architecture in accurately deciphering weak, single-trial electroencephalogram signals amidst complex background noise. Ultimately, these neurophysiological and algorithmic findings lay a solid theoretical and methodological foundation for the objective and quantitative evaluation of individual auditory cognitive capabilities in clinical applications, transcending the fundamental limitations of traditional behavioral paradigms and providing robust technical support for future auditory research and related clinical assessments.

Electroencephalography (EEG) is a non-invasive neurophysiological monitoring technique. It records the electrical activity of the cerebral cortex using electrodes placed on the scalp surface. Owing to its high safety, portability, and millisecond-level temporal resolution, EEG has been widely utilized in a variety of fields, including clinical diagnosis, brain-computer interfaces (BCIs), and cognitive neuroscience research. However, due to its microvolt-level amplitude, EEG is highly susceptible to various artifacts, including electrooculographic (EOG), electrocardiographic (ECG), electromyographic (EMG), and power line interference (PLI). These artifacts can obscure genuine neural activity and introduce spurious electrophysiological features. Consequently, they may compromise EEG signal quality, thereby reducing the reliability of downstream analyses. To address this issue, numerous EEG artifact removal methods have been developed, including both traditional denoising techniques and deep learning-based approaches. Traditional EEG denoising methods have long served as the primary solutions for artifact removal. Representative approaches include filtering, regression, and blind source separation. Although these methods have demonstrated effectiveness in specific scenarios, they suffer from several inherent limitations. Filtering assumes that artifacts and EEG signals can be separated in the frequency domain, but many artifacts, such as EOG and EMG, overlap with EEG spectra, which may lead to the loss of valuable neural information. Regression methods require high-quality artifact references to estimate and subtract contaminations, limiting their effectiveness in reference-free scenarios. Blind source separation can remove artifacts without external references, but it typically requires the number of EEG channels to exceed the number of sources, restricting its application in single- or low-channel EEG recordings. Deep learning-based EEG denoising methods address these limitations effectively. First, they learn the nonlinear mapping between contaminated and clean EEG directly from data in an end-to-end manner. This approach does not rely on assumptions about spectral separability, thereby preserving neural activity more completely. Second, the reference information is incorporated during the training phase, allowing the trained model to perform artifact removal independently without external references. Third, deep learning models can be flexibly designed to accommodate various recording setups, achieving robust denoising for both high-density and single-channel EEG. Collectively, these advantages enable deep learning-based methods to overcome the main challenges of traditional approaches, providing more accurate and reliable EEG signal recovery. The superior denoising performance of deep learning-based EEG denoising methods has attracted increasing attention in EEG artifact removal research. As a result, many deep learning-based denoising methods have been developed and successfully applied in neural engineering areas. However, a systematic review of the techniques and applications in this field is still lacking. To address this gap, this paper reviews recent advances in deep learning-based EEG denoising from four perspectives: technical principle, benchmark dataset, denoising model, and evaluation method. Representative applications in neural signal analysis and BCI decoding are also summarized. Furthermore, the advantage, existing challenge, and future research direction of deep learning-based EEG denoising are discussed. This review aims to provide valuable theoretical insights and technical guidance for researchers. It is also expected to promote further advances and broader applications of deep learning-based EEG denoising techniques.

Objective To evaluate the characteristics, related therapeutic regimens, risk factors of cardiovascular adverse events (CVAEs) and prognosis in the current therapy of newly diagnosed multiple myeloma (NDMM) patients in China. Methods A retrospective observational study was performed on patients with NDMM treated at Zhongshan Hospital, Fudan University, from January 2013 to October 2021. The propensity score matching at a ratio of 1∶1 was employed to balance baseline confounders between the therapeutic group and the control group, and CVAEs incidences were compared between the corresponding groups. Independent risk factors for CVAEs were identified through Fine-Gray competing risk regression model. Kaplan-Meier survival curves and log-rank tests were used to assess overall survival and progression-free survival of patients. Results A total of 829 patients with NDMM were analyzed, with a median of 33.1 months of follow-up for survivors. CVAEs occurred in 65 patients (7.8%) during first-line treatment, and a total of 74 events were recorded, with 52.7% grade 3 or greater in severity. Arrhythmia (particularly atrial fibrillation), thrombotic event (TE), and heart failure were the most common types. The median time from treatment to the first occurrence of CVAEs was 26 days, while the medium occurrence time of TE was later than the other events (76 d vs 19 d, P＝0.003). The exposure to immunomodulatory drugs was related to a higher incidence of TEs, whereas other therapeutic regimens, including bortezomib, didn’t show the effect of cardiovascular toxicity. Results of multivariate analysis showed that Eastern Cooperative Oncology Group performance status ＞2, symptomatic heart diseases within three months before treatment, and estimated glomerular filtration rate≤50 mL·min⁻¹·(1.73 m²)⁻¹ were independent risk factors for CVAEs. Compared with patients without CVAE, patients who experienced CVAEs had significantly inferior progression-free survival and overall survival, and higher cardiovascular mortality (P＜0.001). Conclusions The incidence of CVAE is relatively high during first-line treatment in NDMM patients, and the prognosis of CVAE patients is poor. The risk of CVAE is significantly increased in patients with underlying cardiovascular diseases.

ObjectiveTo investigate the epidemiological characteristics of respiratory adenovirus (AdV) infection among hospitalized patients under 18 years of age with pneumonia in Yancheng City of Jiangsu Province from 2023 to 2024, and to provide a basis for the prevention and control of adenovirus infection. MethodsNasopharyngeal swab samples were collected from hospitalized patients under 18 years of age diagnosed with pneumonia at two national influenza sentinel hospitals in Yancheng City from February 2023 to June 2024. A respiratory pathogen nucleic acid detection kit was used to test for 28 pathogens, including influenza A virus, influenza B virus, respiratory syncytial virus, human rhinovirus, adenovirus, Mycoplasma pneumoniae, Chlamydia pneumoniae, human metapneumovirus, human bocavirus, enterovirus, polyomavirus K1, polyomavirus WU, parainfluenza viruses 1‒4, coronavirus 229E, HKU1, OC43, and NL63, Haemophilus influenzae, Bordetella pertussis, Group A Streptococcus, Staphylococcus aureus, Streptococcus pneumoniae, Legionella pneumophila, Klebsiella pneumoniae, and the novel coronavirus, and to analyze patients’ mixed infections. Adenovirus-positive samples were subjected to polymerase chain reaction (PCR) amplification using Hexon and Fibre as target genes, and the products were analyzed by sequencing. Whole-genome sequencing analyses were performed on positive samples with Ct values less than 22. ResultsAmong 368 hospitalized pneumonia patients under 18 years of age in Yancheng City from February 2023 to July 2024, most were male (51.63%, 190/368), with preschool-aged children (42.66%, 157/368) and primary school-aged children (40.22%, 148/368) comprising the majority. A total of 44 cases tested positive for adenovirus, yielding a positivity rate of 11.96%. Single infections, dual infections, and multiple infections accounted for 61.36% (27/44), 36.36% (16/44), and 2.27% (1/44), respectively. From November 2023, the adenovirus positivity rate rose sharply, peaking at 51.61% in December. Following PCR amplification and sequencing, valid sequences were successfully obtained for 37 strains, comprising 36 AdV-3 strains and 1 AdV-C strain. Eight adenovirus-positive samples with Ct values below 22 were inoculated into HEp-2 cell lines for virus isolation, and whole-genome sequencing analyses were performed on the five successfully isolated strains. Phylogenetic analyses revealed that the AdV-3 strains collectively belonged to the same evolutionary clade. Whole-genome sequencing results indicated that four AdV-3 strains (HAdV-B3) showed no evidence of genetic recombination, whereas one AdV-C strain (HAdV-89) might have undergone genetic recombination. ConclusionThe adenovirus positivity rate among hospitalized pneumonia patients aged under 18 years of age in Yancheng from 2023 to 2024 was slightly higher than that reported in previous studies. Adenovirus infection exhibited a pattern of high incidence in winter, with the AdV-3 type being the predominant pathogen; one AdV-C strain (HAdV-89) might have undergone genetic recombination. It is recommended to strengthen surveillance of respiratory adenovirus type AdV-3 and its variants during the winter season, promote the use of multiplex pathogen detection techniques, and explore the clinical application of an AdV-3 vaccine.

ObjectiveThis study aims to develop and validate a novel multimodal predictive model， termed criss-cross network（CCNet）-directed graph neural network（DGNN）（CGN）， for accurate assessment of pulmonary nodule progression in high-risk individuals for lung cancer， by integrating longitudinal chest computed tomography（CT） imaging with both traditional Chinese and western clinical evaluation data. MethodsA cohort of 4 432 patients with pulmonary nodules was retrospectively analyzed. A twin CCNet was employed to extract spatiotemporal representations from paired sequential CT scans. Structured clinical assessment and imaging-derived features were encoded via a multilayer perceptron， and a similarity-based alignment strategy was adopted to harmonize multimodal imaging features across temporal dimensions. Subsequently， a DGNN was constructed to integrate heterogeneous features， where nodes represented modality-specific embeddings and edges denoted inter-modal information flow. Finally， model optimization was performed using a joint loss function combining cross-entropy and cosine similarity loss， facilitating robust classification of nodule progression status. ResultsThe proposed CGN model demonstrated superior predictive performance on the held-out test set， achieving an area under the receiver operating characteristic curve（AUC） of 0.830， accuracy of 0.843， sensitivity of 0.657， specificity of 0.712， Cohen's Kappa of 0.417， and F₁ score of 0.544. Compared with unimodal baselines， the CGN model yielded a 36%-48% relative improvement in AUC. Ablation studies revealed a 2%-22% increase in AUC when compared to simplified architectures lacking key components， substantiating the efficacy of the proposed multimodal fusion strategy and modular design. Incorporation of traditional Chinese medicine （TCM）-specific symptomatology led to an additional 5% improvement in AUC， underscoring the complementary value of integrating TCM and western clinical data. Through gradient-weighted activation mapping visualization analysis， it was found that the model's attention predominantly focused on nodule regions and effectively captured dynamic associations between clinical data and imaging-derived features. ConclusionThe CGN model， by synergistically combining cross-attention encoding with directed graph-based feature integration， enables effective alignment and fusion of heterogeneous multimodal data. The incorporation of both TCM and western clinical information facilitates complementary feature enrichment， thereby enhancing predictive accuracy for pulmonary nodule progression. This approach holds significant potential for supporting intelligent risk stratification and personalized surveillance strategies in lung cancer prevention.