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.

ObjectiveTo explore the effect and mechanism of Taishan Panshi powder （TSPSP） on inhibiting oxidative stress injury in human chorionic trophoblast cells （HTR-8/SVneo）， and to uelucidate the underlying mechanism of TSPSP in the treatment of spontaneous abortion （SA）. MethodsGene differential analysis of SA was performed using the Gene Expression Omnibus （GEO） database and correlated with oxidative stress. Network pharmacology was employed to screen the active components of TSPSP， and a "Chinese medicine-component-target-disease" network was constructed to predict the mechanism of action of TSPSP. For in vitro validation experiments， HTR-8/SVneo cells were divided into blank group， model group， TSPSP-containing serum 2.5%， 5%， 10% groups， and nuclear factor E₂-related factor 2 （Nrf2） inhibitor group （ML385， 30 μmol·L^-1）. Except for the blank group， other groups were stimulated with 150 μmol·L^-1 H₂O₂ for 3 h to establish a cell oxidative stress injury model. After successful modeling， the blank group and model group were given 10% blank serum， each TSPSP-containing serum group was treated with the corresponding concentration of drug-containing serum， and the Nrf2 inhibitor group was additionally given 30 μmol·L^-1 ML385 on the basis of 10% TSPSP-containing serum. All groups of cells were continuously cultured under the above conditions for 24 h， and then samples were collected for subsequent detection. Cell viability in each group was detected by CCK-8 assay. Cell migration rate was detected by scratch test. The contents of malondialdehyde （MDA）， Fe²⁺， and Glutathione （GSH） were detected by enzyme-linked immunosorbent assay （ELISA）. Intracellular reactive oxygen species （ROS） level was detected by a fluorescent probe （DCF-DA）. The protein and mRNA expression levels of Kelch-like ECH-associated protein 1 （KEAP1）， Nrf2， and forkhead box protein O3 （FoxO3） in cells were detected by immunofluorescence （IF） and real-time quantitative polymerase chain reaction （Real-time PCR）. The protein expression levels of KEAP1， Nrf2， FoxO3， Glutathione peroxidase 4 （GPX4）， and superoxide dismutase （SOD） in cells were detected by Western blot. ResultsThe GSE76862 and GSE22490 datasets were obtained from the GEO database. Differential gene analyses showed that the KEAP1， Nrf2， and FoxO3 genes were all associated with the disease. After matching with the oxidative stress pathway， nine significantly differential pathways were identified （P<0.05）， among which three contained the target genes Nrf2 and FoxO3. A total of 246 active ingredient targets of TSPSP and 2 804 SA-related targets were obtained through network pharmacology， and 154 potential action targets were obtained after taking the intersection. Topological analysis showed that targets such as KEAP1 and Nrf2 exhibited high degree values. GO and KEGG enrichment analyses indicated that the intersection targets were mainly involved in oxidative stress response， FOXO and MAPK signaling pathways， etc. In in vitro experiments， compared with the blank group， the cell viability in the model group was significantly decreased （P<0.01）. Compared with the model group， the cell viability in each TSPSP-containing serum group was significantly increased （P<0.01）. Compared with the 10% TSPSP-containing serum group， the cell viability in the ML385 group decreased to approximately 70% （P<0.01）. Compared with the blank group， the model group showed significantly increased contents of MDA， Fe²⁺， and ROS， decreased GSH expression （P<0.01）， significantly reduced cell migration rate （P<0.01）， and increased protein and mRNA expression levels of KEAP1 and FoxO3 （P<0.01）， while decreased protein and mRNA expression levels of Nrf2， GPX4， and SOD （P<0.01）. Compared with the model group， each TSPSP-containing serum group showed significantly decreased contents of MDA， Fe²⁺， and ROS， increased GSH expression （P<0.01）， significantly increased migration rate （P<0.01）， significantly decreased protein and mRNA expression levels of KEAP1 and FoxO3 （P<0.05， P<0.01）， and significantly increased protein and mRNA expression levels of Nrf2， GPX4， and SOD （P<0.05， P<0.01）. Compared with the 10% TSPSP-containing serum group， the ML385 group showed reversed trends in all indicators （P<0.05， P<0.01）. ConclusionTSPSP can inhibit H₂O₂-induced oxidative stress injury of trophoblast cells， and its mechanism of action may be related to the drug activating the KEAP1/Nrf2/FoxO3 signaling pathway.

ObjectiveTo explore the effect and mechanism of Taishan Panshi powder （TSPSP） on inhibiting oxidative stress injury in human chorionic trophoblast cells （HTR-8/SVneo）， and to uelucidate the underlying mechanism of TSPSP in the treatment of spontaneous abortion （SA）. MethodsGene differential analysis of SA was performed using the Gene Expression Omnibus （GEO） database and correlated with oxidative stress. Network pharmacology was employed to screen the active components of TSPSP， and a "Chinese medicine-component-target-disease" network was constructed to predict the mechanism of action of TSPSP. For in vitro validation experiments， HTR-8/SVneo cells were divided into blank group， model group， TSPSP-containing serum 2.5%， 5%， 10% groups， and nuclear factor E₂-related factor 2 （Nrf2） inhibitor group （ML385， 30 μmol·L^-1）. Except for the blank group， other groups were stimulated with 150 μmol·L^-1 H₂O₂ for 3 h to establish a cell oxidative stress injury model. After successful modeling， the blank group and model group were given 10% blank serum， each TSPSP-containing serum group was treated with the corresponding concentration of drug-containing serum， and the Nrf2 inhibitor group was additionally given 30 μmol·L^-1 ML385 on the basis of 10% TSPSP-containing serum. All groups of cells were continuously cultured under the above conditions for 24 h， and then samples were collected for subsequent detection. Cell viability in each group was detected by CCK-8 assay. Cell migration rate was detected by scratch test. The contents of malondialdehyde （MDA）， Fe²⁺， and Glutathione （GSH） were detected by enzyme-linked immunosorbent assay （ELISA）. Intracellular reactive oxygen species （ROS） level was detected by a fluorescent probe （DCF-DA）. The protein and mRNA expression levels of Kelch-like ECH-associated protein 1 （KEAP1）， Nrf2， and forkhead box protein O3 （FoxO3） in cells were detected by immunofluorescence （IF） and real-time quantitative polymerase chain reaction （Real-time PCR）. The protein expression levels of KEAP1， Nrf2， FoxO3， Glutathione peroxidase 4 （GPX4）， and superoxide dismutase （SOD） in cells were detected by Western blot. ResultsThe GSE76862 and GSE22490 datasets were obtained from the GEO database. Differential gene analyses showed that the KEAP1， Nrf2， and FoxO3 genes were all associated with the disease. After matching with the oxidative stress pathway， nine significantly differential pathways were identified （P<0.05）， among which three contained the target genes Nrf2 and FoxO3. A total of 246 active ingredient targets of TSPSP and 2 804 SA-related targets were obtained through network pharmacology， and 154 potential action targets were obtained after taking the intersection. Topological analysis showed that targets such as KEAP1 and Nrf2 exhibited high degree values. GO and KEGG enrichment analyses indicated that the intersection targets were mainly involved in oxidative stress response， FOXO and MAPK signaling pathways， etc. In in vitro experiments， compared with the blank group， the cell viability in the model group was significantly decreased （P<0.01）. Compared with the model group， the cell viability in each TSPSP-containing serum group was significantly increased （P<0.01）. Compared with the 10% TSPSP-containing serum group， the cell viability in the ML385 group decreased to approximately 70% （P<0.01）. Compared with the blank group， the model group showed significantly increased contents of MDA， Fe²⁺， and ROS， decreased GSH expression （P<0.01）， significantly reduced cell migration rate （P<0.01）， and increased protein and mRNA expression levels of KEAP1 and FoxO3 （P<0.01）， while decreased protein and mRNA expression levels of Nrf2， GPX4， and SOD （P<0.01）. Compared with the model group， each TSPSP-containing serum group showed significantly decreased contents of MDA， Fe²⁺， and ROS， increased GSH expression （P<0.01）， significantly increased migration rate （P<0.01）， significantly decreased protein and mRNA expression levels of KEAP1 and FoxO3 （P<0.05， P<0.01）， and significantly increased protein and mRNA expression levels of Nrf2， GPX4， and SOD （P<0.05， P<0.01）. Compared with the 10% TSPSP-containing serum group， the ML385 group showed reversed trends in all indicators （P<0.05， P<0.01）. ConclusionTSPSP can inhibit H₂O₂-induced oxidative stress injury of trophoblast cells， and its mechanism of action may be related to the drug activating the KEAP1/Nrf2/FoxO3 signaling pathway.

ObjectiveSpinal cord injury (SCI) directly impairs the regulatory function of the autonomic nervous system, induces intestinal dysfunction, and significantly reduces patients’ quality of life. Preclinical studies have shown that electroacupuncture (EA) therapy can regulate the brain-gut axis and is used to treat central nervous system diseases such as major depressive disorder, Alzheimer’s disease and Parkinson’s disease. Recent research has established that fecal microbiota transplantation (FMT) from EA-treated SCI rats restored intestinal motility and colonic morphology. However, it remains unclear whether the regulation of gut microbiota by EA therapy directly contributes to neural repair after SCI. This study aims to explore whether gut microbiota mediates the neuroprotective effect of EA in the treatment of SCI and its possible mechanism. MethodsThe study employed RNA transcriptome analysis of spinal cord tissue to characterize gene expression profiles and to identify key signaling pathways following EA treatment for SCI. Hematoxylin-Eosin (HE) staining and Nissl staining were used to observe the morphological changes in spinal cord tissue. Western blot (WB) and enzyme-linked immunosorbent assay (ELISA) were applied to detect the effects of EA on the expression of proteins related to nucleotide-binding domain leucine-rich repeat and pyrin domain-containing receptor 3 (NLRP3) -dependent pyroptosis. Using 16S rDNA sequencing, the study observed alterations in gut microbiota diversity and community composition in SCI rats. Prior to establishing SCI models, rats were pretreated with an antibiotic cocktail to induce gut dysbiosis, and the effects on intestinal function and spinal cord neural repair were evaluated. FMT was performed to investigate the regulatory effects of post-EA FMT on motor function, general status, liver and spleen indices, and NLRP3-mediated pyroptosis in SCI rats. ResultsEA improved motor function and reduced regulated neuronal cell death in SCI rats. Transcriptomic analysis demonstrated the activation of immune- and inflammation-related pathways post-SCI, including NOD-like receptors, nuclear factor-kappa B(NF-κB), and Toll-like receptor (TLR) pathways. EA primarily influenced intestinal inflammation and autoimmune functions. 16S rDNA sequencing illustrated that EA did not alter the diversity of gut microbiota. However, EA altered the gut microbiota composition in SCI rats, increasing Lactobacillus and Akkermansia genera while rebalancing the Firmicutes/Bacteroidetes ratio. Furthermore, depletion of gut microbiota by antibiotics disrupted the intestinal barrier, reduced the expression of intestinal barrier proteins Zonula Occludens-1 (ZO-1) and Occludin, elevated serum lipopolysaccharide-binding protein (LBP) levels, exacerbated spinal cord tissue damage, and hindered motor function recovery in SCI rats. FMT from donors treated with EA reduced LBP levels in the intestine, blood, and spinal cord of rats, inhibited the TLR4 myeloid differentiation primary response protein 88 (MyD88)-NF‑κB pathway and NLRP3-dependent pyroptosis, and improved motor function. On the other hand, FMT treatment resulted in decreased body weight and food intake, whereas FMT using EA-treated donors effectively alleviated these alterations. ConclusionEA effectively alleviated neuroinflammatory responses in rats with SCI, primarily through regulating the gut microbiota and suppressing the NLRP3-dependent pyroptosis signaling pathway.

ObjectiveSpinal cord injury (SCI) directly impairs the regulatory function of the autonomic nervous system, induces intestinal dysfunction, and significantly reduces patients’ quality of life. Preclinical studies have shown that electroacupuncture (EA) therapy can regulate the brain-gut axis and is used to treat central nervous system diseases such as major depressive disorder, Alzheimer’s disease and Parkinson’s disease. Recent research has established that fecal microbiota transplantation (FMT) from EA-treated SCI rats restored intestinal motility and colonic morphology. However, it remains unclear whether the regulation of gut microbiota by EA therapy directly contributes to neural repair after SCI. This study aims to explore whether gut microbiota mediates the neuroprotective effect of EA in the treatment of SCI and its possible mechanism. MethodsThe study employed RNA transcriptome analysis of spinal cord tissue to characterize gene expression profiles and to identify key signaling pathways following EA treatment for SCI. Hematoxylin-Eosin (HE) staining and Nissl staining were used to observe the morphological changes in spinal cord tissue. Western blot (WB) and enzyme-linked immunosorbent assay (ELISA) were applied to detect the effects of EA on the expression of proteins related to nucleotide-binding domain leucine-rich repeat and pyrin domain-containing receptor 3 (NLRP3) -dependent pyroptosis. Using 16S rDNA sequencing, the study observed alterations in gut microbiota diversity and community composition in SCI rats. Prior to establishing SCI models, rats were pretreated with an antibiotic cocktail to induce gut dysbiosis, and the effects on intestinal function and spinal cord neural repair were evaluated. FMT was performed to investigate the regulatory effects of post-EA FMT on motor function, general status, liver and spleen indices, and NLRP3-mediated pyroptosis in SCI rats. ResultsEA improved motor function and reduced regulated neuronal cell death in SCI rats. Transcriptomic analysis demonstrated the activation of immune- and inflammation-related pathways post-SCI, including NOD-like receptors, nuclear factor-kappa B(NF-κB), and Toll-like receptor (TLR) pathways. EA primarily influenced intestinal inflammation and autoimmune functions. 16S rDNA sequencing illustrated that EA did not alter the diversity of gut microbiota. However, EA altered the gut microbiota composition in SCI rats, increasing Lactobacillus and Akkermansia genera while rebalancing the Firmicutes/Bacteroidetes ratio. Furthermore, depletion of gut microbiota by antibiotics disrupted the intestinal barrier, reduced the expression of intestinal barrier proteins Zonula Occludens-1 (ZO-1) and Occludin, elevated serum lipopolysaccharide-binding protein (LBP) levels, exacerbated spinal cord tissue damage, and hindered motor function recovery in SCI rats. FMT from donors treated with EA reduced LBP levels in the intestine, blood, and spinal cord of rats, inhibited the TLR4 myeloid differentiation primary response protein 88 (MyD88)-NF‑κB pathway and NLRP3-dependent pyroptosis, and improved motor function. On the other hand, FMT treatment resulted in decreased body weight and food intake, whereas FMT using EA-treated donors effectively alleviated these alterations. ConclusionEA effectively alleviated neuroinflammatory responses in rats with SCI, primarily through regulating the gut microbiota and suppressing the NLRP3-dependent pyroptosis signaling pathway.

Objective To investigate the epidemiological characteristics of malaria and provide a basis for developing improved prevention and control measures. Methods Data were obtained from the Chinese Disease Prevention and Control Information System. Malaria surveillance data for Nanning City from January 1,2023, to December 31,2024, were exported from the Infectious Disease Reporting Information Management Subsystem. The characteristics of the two groups of malaria cases were compared. Results A total of 103 imported malaria cases were reported in Nanning City in 2024, representing a 38.32% decrease compared with the same period of the previous year. No statistically significant difference were observed between cases reported in 2023 and 2024 in terms of average age, gender ratio, proportion of parasite species, and monthly reporting distribution;however, statistically significant differences were found in the proportion of reporting areas and current residence areas(χ2= 13.572 and 10.355, respectively; P = 0.001 and 0.035, respectively). The proportion of cases reported in Shanglin County and the proportion of cases residing in Shanglin County were both lower than those during the same period of the previous year. Conclusions The high aggregation of imported malaria cases in Nanning City has decreased. Medical institutions in areas other than Shanglin County should strengthen their vigilance against malaria.

Objective To gain insights into the biological characteristics of different populations of Culex tritaeniorhynchus within and around the Yellow Sea wetland from the perspective of the occurrence of resistance, we investigated the levels of resistance to deltamethrin and kdr gene mutation in the wetland and its peripheral areas. Methods Specimens were collected from Cx. tritaeniorhynchus populations at two monitoring sites in the Rare Bird National Nature Reserve and Tiaozi Ni Wetland Scenic Area, and also from two populations in Yancheng City and the Liuhe District of Nanjing, and the resistance of these mosquitoes to deltamethrin was determined using the CDC biotest bottle method. For each concentration of deltamethrin assessed, a random subset of exposed specimens was selected for amplification of the kdr gene fragment, followed by Sanger sequencing to identify and analyze resistance-associated mutations. Results The LC50 levels of deltamethrin among mosquitoes from the four populations in Luhe, Yancheng, the Rare Bird National Nature Reserve and the Tiaozi Ni Wetland Scenic Area were 2.048 5, 7.798 2, 3.473 3, and 17.695 5 mg/mL, respectively, with corresponding concentrations of deltamethrin ranging from 0.005 to 5.000,0.050 to 50.000,0.050 to 25.000 and 0.050 to 50.000 mg/mL, respectively. Furthermore, the ranges of the KT50 values were 11.76-107.43, 67.05-216.30,29.77-107.43 and 28.40-329.51 min; the 1-h knockdown rates were 34.58%-99.15%, 9.52%-43.80%, 55.09%-73.01%, and 10.09%-68.07%; and the 24-h mortality rates were 12.15%-67.52%,9.52%-79.56%,13.17%-82.21%, and 11.01%-78.99%, respectively. With respect to kdr gene mutation, we assayed a total of 63,70,59, and 57 mosquitoes for the four populations, for which we detected L1014F mutation frequencies of 14.29%, 35.00%, 20.34%, and 31.58%, respectively, with a majority of these mutations being heterozygous for resistance. In addition, five adult mosquitoes were identified has having synonymous mutations at site 1011[i. e. , AAT(asparagine)mutation to AAC(asparagine)]. Conclusions Our findings revealed the clear resistance of Cx. tritaeniorhynchus to deltamethrin in the Yancheng region of the Yellow Sea wetland, and the resistance phenotype and kdr frequency of Cx. tritaeniorhynchus in the wetland environment were comparable to those of Cx. tritaeniorhynchus in the wetland environment, thereby indicating that the resistance of different populations of Cx. tritaeniorhynchus was homogeneous under the pressure of different insecticide selection within and around the wetland. However, the underlying mechanisms need to be further studied.

Objective To evaluate the role of distortion product otoacoustic emissions (DPOAE) testing in evaluating early hearing loss among noise-exposed workers. Methods A total of 174 noise-exposed workers in a metal shipbuilding enterprise were selected as the research subjects by the convenience sampling method. Pure tone audiometry (PTA), DPOAE and the level of noise exposure were conducted on the workers. The rank correlation analysis was used to analyze the correlation between DPOAE amplitude and PTA threshold. The multilevel model was used to analyze the effects of gender, age, noise exposure intensity, cumulative noise exposure (CNE), hearing loss classification and PTA threshold on DPOAE results. Results At the frequencies of 0.50, 1.00, 2.00, 3.00, 4.00, 6.00 and 8.00 kHz, the DPOAE amplitude was negatively correlated with the PTA threshold (rank correlation coefficients were -0.12, -0.48, -0.47, -0.18, -0.23, -0.44, -0.19, respectively, all P<0.01). At the most frequencies, DPOAE amplitude was negatively correlated with age and CNE (all P<0.05). The results of multilevel model analysis showed that there were significant differences in DPOAE amplitudes at certain frequencies across gender, age, noise intensity, CNE, and hearing loss classification (all P<0.05). Significant differences in DPOAE responses were found among different CNE and hearing loss groups (all P<0.01). Conclusion DPOAE testing can objectively reflect the hearing status of noise-exposed workers and could be considered for inclusion in routine hearing monitoring to facilitate early detection of noise-induced hearing loss.

Objective: Huntington’s disease (HD) is characterized by motor, cognitive, and neuropsychiatric symptoms. Oculomotor impairments and gait variability have been independently considered as potential markers in HD. However, an integrated analysis of eye movement and gait is lacking. We performed multiple examinations of eye movement and gait variability in HTT mutation carriers, analyzed the consistency between these parameters and clinical severity, and then examined the associations between oculomotor impairments and gait deficits. Methods: We included 7 patients with pre-HD, 30 patients with HD and 30 age-matched controls. We collected demographic data and assessed the Unified Huntington’s Disease Rating Scale (UHDRS) score. Examinations, including saccades, smooth pursuit tests, and optokinetic (OPK) tests, were performed to evaluate eye movement function. The parameters of gait include stride length, walking velocity, step deviation, step length, and gait phase. Results: HD patients have significant impairments in the latency and velocity of saccades, the gain of smooth pursuit, and the gain and slow phase velocities of OPK tests. Only the speed of saccades significantly differed between pre-HD patients and controls. There are significant impairments in stride length, walking velocity, step length, and gait phase in HD patients. The parameters of eye movement and gait variability in HD patients were consistent with the UHDRS scores. There were significant correlations between eye movement and gait parameters. Conclusion Our results show that eye movement and gait are impaired in HD patients and that the speed of saccades is impaired early in pre-HD. Eye movement and gait abnormalities in HD patients are significantly correlated with clinical disease severity.

In this paper, near-infrared spectroscopy(NIRS) was employed to analyze 129 batches of commercial products of Reduning Injection. The batch reporting rate was estimated according to the report of Reduning Injection in the direct adverse drug reaction(ADR) reporting system of the drug marketing authorization holder of the Center for Drug Reevaluation of the National Medical Products Administration(National Center for ADR Monitoring) from August 2021 to August 2022. According to the batch reporting rate, the samples of Reduning Injection were classified into those with potential risks and those being safe. No processing, random oversampling(ROS), random undersampling(RUS), and synthetic minority over-sampling technique(SMOTE) were then employed to balance the unbalanced data. After the samples were classified according to appropriate sampling methods, competitive adaptive reweighted sampling(CARS), successive projections algorithm(SPA), uninformative variables elimination(UVE), and genetic algorithm(GA) were respectively adopted to screen the features of spectral data. Then, support vector machine(SVM), logistic regression(LR), k-nearest neighbors(KNN), naive bayes(NB), random forest(RF), and artificial neural network(ANN) were adopted to establish the risk prediction models. The effects of the four feature extraction methods on the accuracy of the models were compared. The optimal method was selected, and bayesian optimization was performned to optimize the model parameters to improve the accuracy and robustness of model prediction. To explore the correlations between potential risks of clinical use and quality test data, TreeNet was employed to identify potential quality parameters affecting the clinical safety of Reduning Injection. The results showed that the models established with the SVM, LR, KNN, NB, RF, and ANN algorithms had the F1 scores of 0.85, 0.85, 0.86, 0.80, 0.88, and 0.85 and the accuracy of 88%, 88%, 88%, 85%, 91%, and 88%, respectively, and the prediction time was less than 5 s. The results indicated that the established models were accurate and efficient. Therefore, near infrared spectroscopy combined with machine learning algorithms can quickly predict the potential risks of clinical use of Reduning Injection in batches. Three key quality parameters that may affect clinical safety were identified by TreeNet, which provided a scientific basis for improving the safety standards of Reduning Injection.
Spectroscopy, Near-Infrared/methods* ; Drugs, Chinese Herbal/administration & dosage* ; Machine Learning ; Algorithms ; Humans ; Quality Control