BACKGROUND: Blood glucose and serum albumin have been associated with cardiovascular disease prognosis, but the impact of admission-blood-glucose-to-albumin ratio (AAR) on adverse outcomes in critical ill coronary artery disease (CAD) patients was not investigated. METHODS: Patients diagnosed with CAD were non-consecutively selected from the MIMIC-IV database and categorized into quartiles based on their AAR. The primary outcome was 1-year mortality, and secondary endpoints were in-hospital mortality, acute kidney injury (AKI), and renal replacement therapy (RRT). A restricted cubic splines model and Cox proportional hazard models assessed the association between AAR and adverse outcomes in CAD patients. Kaplan-Meier survival analysis determined differences in endpoints across subgroups. RESULTS: A total of 8360 patients were included. There were 726 patients (8.7%) died in the hospital and 1944 patients (23%) died at 1 year. The incidence of AKI and RRT was 63% and 4.3%, respectively. High AAR was markedly associated with in-hospital mortality (HR = 1.587, P = 0.003), 1-year mortality (HR = 1.502, P < 0.001), AKI incidence (HR = 1.579, P < 0.001), and RRT (HR = 1.640, P < 0.016) in CAD patients in the completely adjusted Cox proportional hazard model. Kaplan-Meier survival analysis noted substantial differences in all endpoints based on AAR quartiles. Stratified analysis and interaction test demonstrated stable correlations between AAR and outcomes. CONCLUSIONS The results highlight that AAR may be a potential indicator for assessing in-hospital mortality, 1-year mortality, and adverse renal prognosis in critical CAD patients.

Azvudine and nirmatrelvir/ritonavir (Paxlovid) are recommended for COVID-19 treatment in China, but their safety and efficacy in the elderly population are not fully known. In this multicenter, retrospective, cohort study, we identified 5131 elderly hospitalized COVID-19 patients from 32,864 COVID-19 patients admitted to nine hospitals in Henan Province, China, from December 5, 2022, to January 31, 2023. The primary outcome was all-cause death, and the secondary outcome was composite disease progression. Propensity score matching (PSM) was performed to control for confounding factors, including demographics, vaccination status, comorbidities, and laboratory tests. After 2:1 PSM, 1786 elderly patients receiving azvudine and 893 elderly patients receiving Paxlovid were included. Kaplan-Meier and Cox regression analyses revealed that compared with Paxlovid group, azvudine could significantly reduce the risk of all-cause death (log-rank P = 0.002; HR: 0.71, 95% CI: 0.573-0.883, P = 0.002), but there was no difference in composite disease progression (log-rank P = 0.52; HR: 1.05, 95% CI: 0.877-1.260, P = 0.588). Four sensitivity analyses verified the robustness of above results. Subgroup analysis suggested that a greater benefit of azvudine over Paxlovid was observed in elderly patients with primary malignant tumors (P for interaction = 0.005, HR: 0.32, 95% CI: 0.18-0.57) compared to patients without primary malignant tumors. Safety analysis revealed that azvudine treatment had a lower incidence of adverse events and higher lymphocyte levels than Paxlovid treatment. In conclusion, azvudine treatment is not inferior to Paxlovid treatment in terms of all-cause death, composite disease progression and adverse events in elderly hospitalized COVID-19 patients.

BACKGROUND:Hypoxia is strongly associated with the development and progression of osteoarthritic chondrocyte injury,but the specific targets and regulatory mechanisms are unclear. OBJECTIVE:A machine learning approach was used to identify KDEL(Lys-Asp-Glu-Leu)receptor 3(KDELR3)as a characteristic gene for osteoarthritis hypoxia and immune infiltration analysis,to provide new ideas and methods for the treatment of osteoarthritis. METHODS:The osteoarthritis-related datasets were downloaded from the GEO database and the GSEA website to obtain hypoxia-related genes.The osteoarthritis datasets were batch-corrected and immune infiltration analyzed using R language,and osteoarthritis hypoxia genes were extracted for differential analysis.Differentially expressed genes were analyzed for GO function and KEGG signaling pathway.Weighted correlation network analysis(WGCNA)and machine learning were also used to screen osteoarthritis hypoxia signature genes,and in vitro cellular experiments were performed to validate expression and correlate immune infiltration analysis using the datasets and qPCR. RESULTS AND CONCLUSION:(1)8492 osteoarthritis genes were obtained by batch correction and principal component analysis,mainly strongly associated with immune cells such as Macrophages M2 and Mast cells resting;200 hypoxia genes were also obtained,resulting in 41 osteoarthritis hypoxia differentially expressed genes.(2)GO analysis involved mainly biological processes such as response to nutrient levels and glucocorticoids;cellular components such as lysosomal lumen and Golgi lumen;and molecular functions such as 14-3-3 protein binding and DNA-binding transcriptional activator activity.(3)KEGG analysis of osteoarthritis hypoxia differentially expressed genes was associated with signaling pathways such as PI3K-Akt,FoxO,and microRNAs in cancer.(4)The characteristic gene KDELR3 was obtained after using WGCNA analysis and machine learning screening.(5)The gene expression of KDELR3 was found to be higher in the test group than in the control group in the synovium(P=0.014)but lower in the meniscus(P=0.024)after validation by gene microarray.(6)In vitro chondrocyte assay showed that the expression of KDELR3 was higher in cartilage than in the control group(P=0.005),while KDELR3 was closely associated with Macrophages M0(P=0.014)and T cells follicular helper(P=0.014).Using a machine learning approach,we confirmed that KDELR3 can be used as a hypoxic signature gene for osteoarthritis and may intervene in osteoarthritis pathogenesis by improving hypoxia,expecting to provide a new direction for better treatment of osteoarthritis.

BACKGROUND:Ferroptosis is strongly associated with the occurrence and progression of osteoarthritis,but the specific characteristic genes and regulatory mechanisms are not known. OBJECTIVE:To identify osteoarthritis ferroptosis signature genes and immune infiltration analysis using the WGCNA and various machine learning methods. METHODS:The osteoarthritis dataset was downloaded from the GEO database and ferroptosis-related genes were obtained from the FerrDb website.R language was used to batch correct the osteoarthritis dataset,extract osteoarthritis ferroptosis genes and perform differential analysis,analyze differentially expressed genes for GO function and KEGG signaling pathway.WGCNA analysis and machine learning(random forest,LASSO regression,and SVM-RFE analysis)were also used to screen osteoarthritis ferroptosis signature genes.The in vitro cell experiments were performed to divide chondrocytes into normal and osteoarthritis model groups.The dataset and qPCR were used to verify expression and correlate immune infiltration analysis. RESULTS AND CONCLUSION:(1)12 548 osteoarthritis genes were obtained by batch correction and PCA analysis,while 484 ferroptosis genes were obtained,resulting in 24 differentially expressed genes of osteoarthritis ferroptosis.(2)GO analysis mainly involved biological processes such as response to oxidative stress and response to organophosphorus,cellular components such as apical and apical plasma membranes,and molecular functions such as heme binding and tetrapyrrole binding.(3)KEGG analysis exhibited that differentially expressed genes of osteoarthritis ferroptosis were related to signaling pathways such as the interleukin 17 signaling pathway and tumor necrosis factor signaling pathway.(4)After using WGCNA analysis and machine learning screening,we obtained the characteristic gene KLF2.After validation by gene microarray,we found that the gene expression of KLF2 was higher in the test group than in the control group in the meniscus(P=0.000 14).(5)In vitro chondrocyte assay showed that type Ⅱ collagen and KLF2 expression was lower in the osteoarthritis group than in the control group in chondrocytes(P<0.05),while in osteoarthritis ferroptosis,mast cells activated was closely correlated with dendritic cells(r=0.99);KLF2 was closely correlated with natural killer cells(r=-1,P=0.017)and T cells follicular helper(r=-1,P=0.017).(6)The findings indicate that using WGCNA analysis and machine learning methods confirmed that KLF2 can be a characteristic gene for osteoarthritis ferroptosis and may improve osteoarthritis ferroptosis by interfering with KLF2.

In this study, doxorubicin (DOX) was used as the model drug, new indocyanine green (IR820) as the photosensitizer, and temperature sensitive liposomes (TSL) as the carrier. H460-NCI photoheat-sensitive liposomes coated with cell membrane of human cell lung cancer (DOX-IR820-TSL@CCM) for highly effective multi-pathway tumor targeting in chemical-photothermal therapy and photodynamic therapy. DOX-IR820-TSL was prepared by reverse evaporation, cancer cell membrane (CCM) was prepared by lysis, crushing and centrifugation, and DOX-IR820-TSL@CCM was prepared by nanomembrane extrusion. The drug-loading conditions of DOX-IR820-TSL were finally determined: the ratio of organic phase to aqueous phase was 4.02, the dosage of dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) was 10.04 mg, and the lipid ratio was 0.12, and the significant phase transition temperature (T_m) of DOX-IR820-TSL was 43.05 ℃. The average particle size of DOX-IR820-TSL@CCM was 153.4 nm, the PDI was 0.279, and the zeta potential was -26.2 mV. The transmission electron microscope (TEM) image shows a homogeneous spherical structure and a translucent film layer. Under near-infrared irradiation, the drug release rate reaches 63.98%, which has adjustable photothermal conversion capacity and the ability to generate reactive oxygen species. Through SDS-PAGE electrophoresis, Western blot, cytotoxicity experiments and cell uptake experiments, it was proved that the design of cell membrane coating can well retain CD47, N-cadherin, CD44, CD326 and other related functional proteins, so that DOX-IR820-TSL@CCM has good immune evasion, homologous adhesion and homologous targeting. In this paper, DOX-IR820-TSL@CCM with camouflage properties and tumor targeting properties were successfully prepared, which can be used as a promising synergistic therapeutic diagnostic platform for future lung cancer treatment. All animal research programs have been approved by the Animal Ethics Committee of Heilongjiang University of Chinese Medicine (number: 2022121011).

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.