Objective To establish an ultrasound-assisted nanonization process for realgar and elucidate the scientific connotation of achieving toxicity reduction and efficacy enhancement within safety thresholds through nanonization. Methods The parameters of zirconium ball water-grinding were optimized using central composite design-response surface methodology (CCD-RSM), and nano-realgar was prepared by combining acid washing with ultrasonic shearing. Four processed products (dry-ground Realgar, traditional water-floating Realgar, SDS-assisted water-floating Realgar and nano-Realgar) were characterized by laser particle size analyzer, scanning electron microscope (SEM), specific surface area and porosity analyzer based on the Brunauer-Emmett-Teller (BET) method, and liquid chromatography-atomic fluorescence spectrometry (LC-AFS). Toxicity was evaluated using a Caenorhabditis elegans model, and efficacy was assessed using a mouse wound model. Results The optimal process parameters of zirconium ball water-grinding were determined as milling for 3.9 h, three cycles of levigation, and the addition of 190 g of zirconia balls (the mass ratio of zirconium balls with diameters of 1, 2, 3, 4 mm is 1∶1∶2∶4); The particle size of Realgar sample can be reduced to 300 nm 30 min of sonication at 250 W. BET analysis showed that nano-realgar had well-developed mesopores. LC-AFS results indicated that, compared with the dry-milled product, all processing methods reduced the As³⁺ content, with nano-Realgar being slightly higher but still below the safety limit. In the C. elegans assay, nano-Realgar exhibited the lowest toxicity (P < 0.001). In the mouse wound model, it achieved the highest healing rate, significantly reducing inflammatory factors and upregulating growth factor expression (P < 0.05). Conclusion The zirconium ball water-grinding combined with ultrasound-assisted nanonization process successfully prepares nano-Realgar, which possesses dual advantages of toxicity reduction and efficacy enhancement within safety thresholds, providing a theoretical basis for upgrading the processing technology of Realgar.

Objective To elucidate the molecular mechanism by which Lobelia chinensis-derived extracellular vesicle-like nanoparticles (LC-EVLP) induce lethal autophagy in hepatocellular carcinoma (HCC) by remodeling the glutamine metabolic network, and to systematically evaluate their antitumor efficacy and biosafety. Methods LC-EVLP were isolated by differential ultracentrifugation and physicochemically characterized. By integrating network pharmacology, molecular docking and multidimensional cell-based assays, the effects of LC-EVLP on the proliferation, migration and invasion of human HCC Mahlavu cells were assessed. Rescue experiments using an autophagic-flux inhibitor and the metabolic intermediate α-ketoglutarate (α-KG) were conducted to interrogate the “metabolism-autophagy” regulatory axis. A nude mouse subcutaneous xenograft model was established to evaluate the in vivo anti-tumor effects of LC-EVLP and histopathological changes in major organs. Results LC-EVLP was successfully isolated and characterized as typical vesicular structures with particle sizes mainly distributed between 126.0 and 151.7 nm, and were efficiently internalized by Mahlavu cells. Functional assays demonstrated that LC-EVLP significantly reduced the viability of Mahlavu cells at concentrations ≥ 12.5 μg/mL (P < 0.001), while exerting no significant cytotoxicity toward LO2 cells. Further analyses revealed that LC-EVLP significantly inhibited colony formation and DNA synthesis (P < 0.05, 0.01, 0.001), and significantly suppressed cell migration and invasion (P < 0.001). The autophagy inhibitor bafilomycin A1 (BafA1) significantly rescued cell viability (P < 0.001), indicating that autophagic cell death was the dominant mechanism underlying the anti-HCC activity of LC-EVLP. LC-EVLP significantly upregulated Beclin 1 expression (P < 0.05, 0.01, 0.001), increased microtubule associated protein 1 light chain 3-II/I (LC3-II/I) value (P < 0.05, 0.01), and reduced p62 protein expression (P < 0.01, 0.001). Moreover, BafA1 further enhanced LC3-II/I accumulation and reversed the decrease in p62 expression (P < 0.01, 0.001), confirming the promotion of a complete autophagic flux. In terms of metabolic regulation, LC-EVLP significantly downregulated the expressions of pyruvate dehydrogenase kinase 1 (PDK1), pyruvate kinase M (PKM) and glutaminase 1 (GLS1) (P < 0.01, 0.001), while α-KG supplementation reversed the abnormal autophagic phenotype (P < 0.01). The in vivo experimental results showed that LC-EVLP significantly inhibited tumor growth (P < 0.05, 0.001), downregulated Ki67 and PCNA expressions (P < 0.001), and induced changes in autophagy related proteins (P < 0.05, 0.01, 0.001), with no significant toxicity observed. Conclusion LC-EVLP exerts anti-tumor effects by regulating PDK1/PKM axis, disrupting cellular metabolic balance and inducing enhanced autophagy. Its low toxicity characteristics suggest its potential for further research as a natural nanomaterial-based antitumor agent.

Objective To study the effect of gamabufotalin (CS-6) on the proliferation of nasopharyngeal carcinoma cells, and explore the mechanism by which CS-6 inhibits nasopharyngeal carcinoma cell proliferation through cell cycle arrest and autophagy. Methods The inhibitory effect of CS-6 on the proliferation of HK-1 and 5-8F cells was detected through CCK-8 assay and colony formation assay. Flow cytometry was used to detect the effect of CS-6 on the cell cycle distribution and apoptosis of HK-1 and 5-8F cells. Western blotting was used to detect the expressions of Cyclin B1 and cyclin dependent protein kinase 1 (CDK1). Transcriptome sequencing was used to screen for differentially expressed genes after CS-6 treatment, and IPA software was used to analyze the enrichment pathways of differentially expressed genes. The ultrastructure of autophagosomes after CS-6 acts on cells was observed using transmission electron microscopy. Western blotting was used to detect the expressions of microtubule associated protein light chain 3 (LC3), p62, and glycogen synthase kinase-3β (GSK-3β)/mammalian target of rapamycin (mTOR) pathway related proteins. After transient transfection of mCherry-EGFP-LC3 plasmid, the changes in autophagy flux of cells treated with CS-6 were observed using laser confocal microscopy. In order to investigate the effect of PFKFB4 on CS-6-induced proliferation, cell cycle arrest and autophagy of nasopharyngeal carcinoma cells, HK-1 and 5-8F cells were transfected with PFKFB4 plasmid, qRT-PCR and Western blotting were used to detect the expression of 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase-4 (PFKFB4), as well as changes in cell survival rate, colony formation, autophagy flux, cell cycle and autophagy related protein expressions. A mouse model of nasopharyngeal carcinoma transplantation in vivo was established, after CS-6 administration, tumor volume and weight were measured, pathological changes in tumor, liver and kidney tissues were observed using hematoxylin-eosin staining, Ki67 expression in tumor tissues was detected using immunohistochemistry, LC3, p62, Cyclin B1, CDK1, p-mTOR, PFKFB4 and p-GSK-3β protein expressions in tumor tissues were detected using Western blotting. Results After 48 h of treatment, CS-6 significantly inhibited the proliferation of HK-1 and 5-8F cells (P < 0.05, 0.001), with half-maximal inhibitory concentrations of 8.684 and 14.320 nmol/L, respectively. CS-6 caused significant G₂/M phase arrest in HK-1 and 5-8F cells (P < 0.05, 0.01), while increasing protein expressions of Cyclin B1 and CDK1 (P < 0.05, 0.01, 0.001). CS-6 did not induce apoptosis in HK-1 and 5-8F cells under treatment conditions of 48 h. CS-6 significantly reduced the mRNA and protein expressions of PFKFB4 (P < 0.05, 0.01, 0.001), typical double-membrane autophagosomes and autolysosomes were observed in the cells, CS-6 upregulated LC3-II/I and downregulated p62 protein expressions (P < 0.05, 0.01, 0.001), promoted the fusion of autophagosomes and lysosomes. Overexpression of PFKFB4 could partially reverse the inhibitory effect of CS-6 on cell proliferation and regulate the expressions of cell cycle and autophagy related proteins (P < 0.05, 0.01, 0.001), and inhibit the fusion of autophagosomes and lysosomes. CS-6 significantly increased the expression of p-GSK-3β and inhibited the expression of p-mTOR (P < 0.05, 0.01, 0.001). Overexpression of PFKFB4 could partially reverse the regulatory effects of CS-6 on p-GSK-3β and p-mTOR (P < 0.05, 0.01). In vivo, CS-6 significantly inhibited tumor growth in tumor bearing mice (P < 0.001), induced pathological damage to tumor tissue, inhibited tumor cell proliferation (P < 0.001), upregulated LC3-II/I, p-GSK-3β expressions and downregulated p62 expression in tumor tissue (P < 0.001), downregulated protein expressions of p-mTOR, PFKFB4, Cyclin B1 and CDK1 (P < 0.01, 0.001), and had no effect on liver and kidney tissues. Conclusion CS-6 exerts an anti-nasopharyngeal carcinoma effect by downregulating PFKFB4, thereby regulating GSK-3β/mTOR signaling pathway and inducing autophagy and cell-cycle arrest.

Objective To elucidate the material basis of Quzhou Chenpi (Citri Reticulatae Pericarpium) extract (CPE) and to investigate its mechanism in improving functional dyspepsia (FD) by repairing the intestinal barrier. Methods CPE was prepared by water decoction combined with freeze-drying. Qualitative and quantitative analyses of its chemical components were performed using ultra-performance liquid chromatography quadrupole time-of-flight mass spectrometry (UPLC-Q-TOF-MS) and high-performance liquid chromatography (HPLC), followed by network pharmacology to predict potential targets and pathways. In vivo, an FD rat model was established by iodoacetamide combined with irregular feeding. After drug intervention, gastric emptying rate and small intestinal propulsion rate were measured. Histopathological changes in the stomach and duodenum were examined by hematoxylin-eosin staining. Levels of lipopolysaccharide (LPS), diamine oxidase (DAO) in serum as well as tumor necrosis factor-α (TNF-α) in duodenal tissue were determined by ELISA. The expressions of zonula occludens-1 (ZO-1) and Occludin in duodenal tissue was assessed by immunohistochemistry. In vitro, a TNF-α-induced Caco-2 intestinal barrier injury model was used. Transepithelial electrical resistance (TEER) value and fluorescein isothiocyanate-dextran with relative molecular mass 4 × 10⁴ (FD-4) permeability were measured. The distribution of ZO-1 and Occludin proteins was observed by immunofluorescence staining. The mRNA expressions of ZO-1, Occludin, interleukin-6 (IL-6) and TNF-α were detected by qRT-PCR. IL-6 secretion was measured by ELISA. The phosphorylation levels of cellular Src (c-Src) and protein kinase B (Akt) were detected by Western blotting. Results A total of 84 chemical components were identified from CPE, eight of which were reported in Citri Reticulatae Pericarpium for the first time. The contents of the four major flavonoids were ranked from high to low as hesperidin > narirutin > nobiletin > tangeretin. Network pharmacology predicted that the potential core targets of CPE for improving FD involved SRC, AKT1, etc., with enrichment in phosphoinositide 3-kinase (PI3K)/Akt signaling pathway and tight junction regulation. In vivo, CPE significantly ameliorated gastrointestinal motility disorders in FD rats (P < 0.05, 0.01, 0.001), alleviated duodenal mucosal injury, reduced levels of LPS and DAO in serum (P < 0.01, 0.001), suppressed TNF-α protein and mRNA expressions in duodenal tissue (P < 0.01, 0.001), and up-regulated Occludin protein expression (P < 0.05, 0.001). In vitro, the actual hesperidin content in CPE was only 10.48% of that in the hesperidin monomer group, yet CPE was equivalent to hesperidin monomer in stabilizing TEER and reducing FD-4 permeability (P < 0.01, 0.001). CPE significantly up-regulated the mRNA expressions of ZO-1 and Occludin (P < 0.001), maintained their continuous network-like distribution, significantly inhibited IL-6 transcription and secretion (P < 0.01, 0.001), and significantly reduced the values of p-c-Src/c-Src and p-Akt/Akt (P < 0.05). Conclusion CPE significantly ameliorates gastrointestinal motility disorders and duodenal mucosal barrier injury in FD rats, and alleviates barrier dysfunction and inflammation in TNF-α-induced Caco-2 cells. Its effect depends on the inhibition of over-activation of c-Src/Akt signaling pathway. The overall efficacy demonstrates the multi-component synergistic characteristics of traditional Chinese medicine, providing an experimental basis for the application of Quzhou Citri Reticulatae Pericarpium in gastrointestinal diseases.

Objective To investigate whether Yiqi Fumai Lyophilized Injection [注射用益气复脉（冻干）, YQFM] ameliorates myocardial hypertrophy and pressure overload-induced heart failure (HF) by upregulating the cardiac endothelial mechanosensitive ion channel Piezo1 and activating the nitric oxide (NO) signaling pathway. Methods A mouse HF model was prepared via transverse aortic constriction (TAC) surgery. Two weeks post-operation, mice received intraperitoneal injections of different doses of YQFM (0.13, 0.26, 0.53 g/kg) or captopril (20 mg/kg) for four weeks. Cardiac function was assessed by echocardiography. Pathological injury and fibrosis were evaluated by hematoxylin and eosin (HE) and Masson staining. Cardiac remodeling and cardiomyocyte hypertrophy were assessed by heart index, heart weight-to-tibia length ratio and wheat germ agglutinin (WGA) staining. Serum biochemical assays, qRT-PCR, Western blotting and immunofluorescence were employed to detect HF-related biomarkers and the expressions of key proteins in Piezo1/NO pathway. Results YQFM significantly improved TAC-induced cardiac dysfunction and remodeling (P < 0.05, 0.01), and alleviated histopathological disorganization, cardiomyocyte hypertrophy and myocardial fibrosis (P < 0.01). Mechanistic studies showed that YQFM reduced levels of cardiac injury markers cardiac troponin T (cTn-T) and N-terminal pro-B-type natriuretic peptide (NT-proBNP) in serum (P < 0.01), and significantly downregulated the expressions of hypertrophy-related genes natriuretic peptide A (Nppa), natriuretic peptide B (Nppb), myosin heavy chain 7 (Myh7) and fibrosis-related genes collagen type I alpha 1 chain (Col1a1), collagen type III alpha 1 chain (Col3a1) in cardiac tissue (P < 0.05, 0.01). YQFM significantly decreased the levels of follistatin-like 1 (Fstl1) and endothelin-1 (ET-1) (P < 0.05, 0.01), while elevating the levels of NO in serum and cardiac tissue and cyclic guanosine monophosphate (cGMP) in cardiac tissue (P < 0.01). Furthermore, YQFM significantly upregulated the expressions of Piezo1 protein and phosphorylated endothelial nitric oxide synthase (p-eNOS) in cardiac endothelial cells (P < 0.01). Conclusion YQFM improves cardiac function and remodeling, and inhibits cardiomyocyte hypertrophy in HF by upregulating endothelial Piezo1 expression and activating eNOS/NO signaling pathway.

Objective To explore the metabolic regulatory effects of different administration methods of brucine-loaded microneedles on rats with collagen-induced arthritis (CIA) through synovial fluid and serum metabolomics, and to provide experimental evidence for elucidating the mechanism of brucine-loaded microneedles in the treatment of rheumatoid arthritis (RA).Methods The non-targeted metabolomics method of ultra-high performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) was adopted. The control group, brucine-loaded hydrogel acupoint microneedles group (XWMN), brucine-loaded hydrogel non-acupoint microneedles group (FXWMN), blank microneedles group (KBMN) and traditional acupuncture group (ZC) were respectively compared with model group. The changes in the profiles of endogenous metabolites in the synovial fluid and serum of CIA rats after intervention were systematically analyzed.Results Based on the screening of synovial fluid metabolomics, differential metabolites were obtained for comparison between each treatment group and the model group. Analysis of the reversal trends of the differential metabolites showed that, compared with model group, three metabolites in XWMN group, five in FXWMN group, three in KBMN group, and three in ZC group were significantly restored toward normal levels. The metabolic regulation of XWMN group mainly involved the cyclic adenosine monophosphate (cAMP) signaling pathway, butyrate metabolism, nucleotide metabolism and other pathways, with a higher number of enriched pathways compared to other treatment groups. Based on serum metabolomics screening, differential metabolites were obtained for comparison between each treatment group and the model group. Through the analysis of the regression trend of differential metabolites, it was found that compared with model group, five metabolites in XWMN group showed significant regression, 15 metabolites in FXWMN group showed significant regression, 12 metabolites in KBMN group showed significant regression, and three metabolites in ZC group showed significant regression; The metabolic regulation of XWMN group mainly involved key pathogenic pathways such as purine metabolism and arachidonic acid metabolism in RA, while FXWMN group involved three pathways including purine metabolism and ABC transporter pathway, and KBMN group involves seven metabolic pathways including purine metabolism and linoleic acid metabolism. Conclusion Different administration methods of brucine-loaded microneedles can modulate the aberrant metabolism in CIA rats to varying degrees. Notably, acupoint administration of brucine-loaded microneedles exerts anti-inflammatory effects and regulates immune cell differentiation by modulating the expression of metabolites such as 3-hydroxybutyrate and deoxycytidine in synovial fluid, thereby influencing pathways like cAMP signaling, butanoate metabolism, and nucleotide metabolism. Concurrently, it exerts anti-inflammatory effects by modulating serum metabolic pathways including purine metabolism and arachidonic acid metabolism. Compared with non-acupoint administration, blank-microneedle treatment and traditional acupuncture, acupoint administration demonstrates a more prominent modulatory effect on key pathogenic pathways (e.g., purine metabolism, arachidonic acid metabolism).

Objective To explore the effect of total saponins of Platycodon grandiflorum on cartilage injury and Ras homolog gene family member A (RhoA)/Rho-associated coiled-coil kinase (ROCK) signaling pathway in knee osteoarthritis (KOA) rats.Methods Knee joint injection of papain was used to prepare a KOA model. The rats were randomly divided into model group, glucosamine sulfate (0.2 g/kg) group, total saponins of P. grandiflorum low- and high-dose (100, 200 mg/kg) groups and total saponins of P. grandiflorum (200 mg/kg) + RhoA activator LPA (1 mg/kg) group. A sham group was also included, with 15 rats in each group. After four weeks of continuous administration, micro CT scanning was used to observe changes in knee joints of rats in each group. Hematoxylin-eosin (HE) and Safranin O-fast green staining were used to observe the pathological changes of cartilage tissue. ELISA was used to detect the levels of matrix metalloproteinase-13 (MMP-13), tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) in joint fluid. TUNEL staining was used to observe chondrocyte apoptosis, and transmission electron microscopy was used to observe the microstructure of chondrocytes. Western blotting was used to detect expressions of aggrecan, type II collagen (Col II), B-cell lymphoma-2 (Bcl-2), Bcl-2 associated X protein (Bax), cleaved cystein-asparate protease-3 (cleaved Caspase-3), RhoA, ROCK1 and ROCK2 proteins in cartilage tissue. Results Compared with sham group, rats in model group showed severe pathological damage to the cartilage tissue, destruction of the ultrastructure of chondrocytes, Mankin score, osteoarthritis research society international (OARSI) score, chondrocyte apoptosis rate, levels of TNF-α, MMP-13, IL-6 in joint fluid, as well as protein expression levels of RhoA, ROCK1, ROCK2, Bax and cleaved Caspase-3 in cartilage tissue were significantly increased (P < 0.05), bone volume/total volume (BV/TV), trabecular thickness (Tb.Th) and protein expression levels of Bcl-2, aggrecan, Col-II in cartilage tissue were significantly reduced (P < 0.05). Compared with model group, glucosamine sulfate group and total saponins of P. grandiflorum groups showed reduced pathological damage to cartilage tissue, normal ultrastructure of chondrocytes, Mankin score, OARSI score, chondrocyte apoptosis rate, levels of TNF-α, MMP-13, IL-6 in joint fluid, as well as protein expression levels of RhoA, ROCK1, ROCK2, Bax and cleaved Caspase-3 in cartilage tissue were significantly decreased (P < 0.05). BV/TV, Tb.Th and protein expression levels of Bcl-2, aggrecan, Col-II in cartilage tissue were significantly increased (P < 0.05). After the combined use of RhoA activator LPA, the improvement effect of total saponins of P. grandiflorum on the above indicators was partially reversed (P < 0.05). Conclusion Total saponins of P. grandiflorum can improve cartilage damage in KOA rats by inhibiting RhoA/ROCK pathway.

Objective To construct HerbVLM, a vision-text fusion model for the identification of traditional Chinese medicinal materials, aiming to improve intelligent recognition performance in distinguishing visually similar categories and generalizing across scenarios. Methods Based on a vision-language pretrained dual-encoder framework, a traditional Chinese medicine knowledge-enhanced prompt module (TKPM) was introduced on the text side, while a hierarchical image adaptation and de-confusion module (HIA) was constructed on the image side. Dynamic fusion and joint loss optimization were further incorporated. Comparative and ablation experiments were conducted on the open-source Chinese-Medicine-163 dataset, and cross-domain and few-shot adaptation validation was performed on a self-collected external dataset of the “new Zhebawei”. Results On the test set, HerbVLM achieved top-1 accuracy (Top-1 Acc), macro precision (P_macro), macro recall (R_macro), and macro F1-score (F1_macro) of 98.69%, 98.70%, 98.71%, and 98.70%, respectively, outperforming comparative methods including residual network (ResNet), context optimization (CoOp), conditional context optimization (CoCoOp), training-free CLIP-Adapter (Tip-Adapter), multi-modal prompt learning (MaPLe), and contrastive vision-language pretraining in Chinese (Chinese-CLIP). Ablation experiments showed that text enhancement, image adaptation, de-confusion constraints, and dynamic fusion all brought stable performance gains. In external testing on the self-collected Zhejiang Eight Flavors dataset, the recognition rates for known categories Lingzhi (Ganoderma), Fupenzi (Rubi Fructus), Tiepishihu (Dendrobii Officinalis Caulis), and Qianhu (Peucedani Radix) were 96.00%, 100.00%, 100.00%, and 100.00%, respectively. For unknown categories, the average accuracies under 1-shot, 2-shot, and 5-shot settings were approximately 84.00%, 91.54%, and 98.20%, respectively, demonstrating favorable few-shot transfer capability. Conclusion HerbVLM significantly improves the fine-grained recognition accuracy and cross-domain generalization performance of traditional Chinese medicinal materials through multimodal collaborative modeling of image-text knowledge, and can provide methodological support for traditional Chinese medicinal material teaching, quality inspection assistance.

Objective To address the strong coupling, long time delays, and nonlinear characteristics inherent in the traditional Chinese medicine (TCM) extraction process, as well as the issues of prediction lag and extreme-value smoothing commonly exhibited by traditional deep time-series models during operating condition transitions, a decoupled digital twin modeling method of key process parameters for the TCM extraction process, integrating Standard Operating Procedure (SOP) priors, named SOP-guided channel-independent patch time series Transformer (SGC-PatchTST), is proposed. Methods Based on the PatchTST architecture, the proposed model introduces three major enhancements: 1. An orthogonal decoupling late-fusion mechanism is employed to transform discrete SOP instructions into process-intention features, utilizing a soft-gating mechanism for fusion to suppress the lag caused by operating condition switches; 2. A decoupled structure comprising primary and auxiliary branches is constructed to independently model the reflux temperature, thereby mitigating the interference caused by multivariable coupling; 3. An extrema-sensitive hybrid loss function is designed, combining local statistics with extreme-value penalties to enhance the model's extrema-fitting and noise-resistance capabilities under the low-pass filtering conditions of Patch-based tokenization. All baseline models uniformly adopt the OneCycleLR learning rate schedule and an early-stopping strategy. Experimental data are independently partitioned into training, validation, and test sets on a batch-by-batch basis to strictly eliminate cross-batch data leakage. Actual production data from a TCM extraction workshop are utilized to construct the predictive model and evaluate it via autoregressive multi-step inference. Results Under a 50-batch stratified split, the single-evaluation Mean Absolute Errors (MAEs) of the upper and lower temperatures for SGC-PatchTST (optimized hyperparameters τ＝1.5σ，λ＝2.0) decrease to 0.36 ℃ and 0.35 ℃, respectively (which are 40.0% and 50.2% lower than the standard PatchTST model, respectively). Its capability to resist extreme deviations is significantly enhanced, with the 95% percentile errors (P95 errors) for the upper and lower temperatures controlled at 1.31 ℃ and 1.33 ℃, respectively. At the same time, benefiting from the decoupled primary and secondary architectures, the MAE of the reflux temperature, which is susceptible to external disturbances, is reduced to a lower value of 0.04 ℃. The prediction of minor fluctuations in the core safety parameter (in-tank pressure) is also maintained at a relatively high precision of 0.000 6 MPa. Furthermore, under three independent stratified splits, the average upper and lower temperature MAEs remain at 0.33 ℃ (coefficient of variation, CV = 6.8%) and 0.32 ℃ (CV = 6.4%), respectively, confirming its good statistical robustness. A six-step combination ablation study and a 25-group hyperparameter grid search validate the contribution of each module and the hyperparameter robustness, respectively. Conclusion The proposed SGC-PatchTST method achieves an ideal level of prediction accuracy and timeliness while demonstrating strong physical plausibility, providing reliable support for subsequent model-based reinforcement learning control.

Objective To explore the formulation rules of existing sleep-improving health foods, construct “data mining statistics-medicine and edible homology/classical formula screening-traditional Chinese medicine theoretical interpretation” new formula system and investigate its potential mechanisms of action. Methods The publicly available record information of sleep-improving health foods was collected from the State Administration for Market Regulation, and the raw materials and compatibility rules were analyzed. Based on the data mining results, a novel sleep-improving formula of medicine and edible homology was constructed. Network pharmacology was used to predict the potential action targets and signaling pathways, and molecular docking was performed to verify the binding affinity between active components and core targets. Results A total of 541 sleep-improving health foods meeting the inclusion criteria were included, involving 154 Chinese medicinal materials. The high-frequency materials included Suanzaoren (Ziziphi Spinosae Semen), Wuweizi (Schisandrae Chinensis Fructus), and Fuling (Poria). Their nature and taste are mainly sweet and neutral, and its meridians mainly focus on the heart and liver meridians. Association rule analysis revealed core drug pairs such as “Schisandrae Chinensis Fructus-Ziziphi Spinosae Semen” and “Poria-Ziziphi Spinosae Semen”. Capsules, tablets, and granules were the main dosage forms. The usage and dosage were mainly 1–2 times daily, and 99.8% of the products did not specify a course of administration. Besides the core function of sleep improvement, these products often also claimed enhancing immunity and relieving physical fatigue as combined functions. The suitable population was primarily individuals with poor sleep status, while the unsuitable population mainly included children, pregnant and lactating women. A novel formula consisting of Baihe (Lilii Bulbus), Dihuang (Rehmanniae Radix), Ziziphi Spinosae Semen, Tianma (Gastrodiae Rhizoma), Gegen (Puerariae Lobatae Radix) was constructed. Network pharmacology predicted 42 active components, 760 action targets, among which 189 were associated with sleep, 196 with immune enhancement, and 63 with gut microbiota regulation. The enrichment analysis indicated that the main pathways involved neuroactive ligand-receptor interaction, phosphatidylinositol-3-hydroxykinase (PI3K)-protein kinase B (Akt) signaling pathway, and endocrine resistance, which are closely related to sleep improvement, immune enhancement, and gut microbiota regulation. Molecular docking showed that the active components of the novel formula bound well with the core targets. Conclusion A new formula of health food for improving sleep, “Lilii Bulbus-Rehmanniae Radix-Ziziphi Spinosae Semen-Gastrodiae Rhizoma-Puerariae Lobatae Radix”, is screened out. Its sleep-improving effect is mainly closely related to enhancing immunity and regulating intestinal microbiota, which provides scientific theoretical support for the subsequent development of health products for sleep improvement..