ObjectiveTo elucidate the efficacy connotation of Shudi Qiangjin pills （SQP） against liver and kidney deficiency in steroid-induced osteonecrosis of femoral head （SONFH） from the perspective of the "disease-syndrome-formula" association and to clarify the underlying mechanisms based on in vivo and in vitro experiment validation. MethodsThe chemical components and the corresponding putative targets of SQP were collected from the Integrative Pharmacology-based Research Platform of Traditional Chinese Medicine （TCMIP） v2.0， the Encyclopedia of Traditional Chinese Medicine （ETCM） v2.0， and HERB databases. The SONFH-related genes were identified based on the differential expression profiles of peripheral blood of patients with SONFH compared to the healthy volunteers， and the disease phenotype-related targets were collected from the TCMIP v2.0 database. Then， the interaction network of "SONFH-related genes and candidate targets of SQP" was constructed based on "gene-gene interaction information"， and the major network targets were screened by calculating the topological characteristic values of the network followed by the functional mining according to the Kyoto Encyclopedia of Genes and Genomes （KEGG） database and the SoFDA database. After that， the SONFH rat model was prepared by lipopolysaccharide combined with methylprednisolone injection， and 2.5， 5， 7.5 g·kg^-1 SQP （once per day， equivalent to 1， 2， and 3 times the clinical equivalent dose， respectively） or 7.3×10^-3 g·kg^-1 of alendronate sodium （ALS， once per week， equivalent to the clinical equivalent dose） was given for 8 weeks. The effect characteristics of SQP and ALS in the treatment of SONFH were evaluated by micro-computed tomography scanning， hematoxylin and eosin staining， alkaline phosphatase （ALP） staining， immunohistochemical staining， enzyme-linked immunosorbent assay， and terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling（TUNEL）staining， and a comparative efficacy analysis was conducted with ALS. In addition， SONFH cell models were prepared by dexamethasone stimulation of osteoblasts， and the intervention was carried out with the medicated serum of SQP at the aforementioned three doses. Cell counting kit-8， ALP staining， ALP activity assay， alizarin red staining， and flow cytometry were employed to investigate the regulatory effect of SQP on osteoblasts. The expression levels of osteogenesis-related proteins and key factors of the target signaling axis were detected by quantitative real-time polymerase chain reaction and Western blot. ResultsThe network analysis results demonstrated that the candidate targets of SQP primarily exerted their therapeutic effects through key signaling pathways， including phosphoinositide 3-kinase（PI3K）/protein kinase B（Akt）， lipid metabolism and atherosclerosis， prolactin， chemokines， and neurotrophic factors pathways. These pathways were significantly involved in critical biological processes such as muscle and bone metabolism and the regulation of the "neuro-endocrine-immune" network， thereby addressing both modern medical symptoms （e.g.， delayed skeletal maturation and recurrent fractures） and traditional Chinese medicine （TCM） symptoms （e.g.， fatigue， aversion to cold， cold limbs， and pain in the limbs and joints in patients with SONFH characterized by liver and kidney deficiency syndrome. Among these pathways， the PI3K/Akt signaling pathway exhibited the highest degree of enrichment. The in vivo experimental results demonstrated that starting from the 4^th week after modeling， the modeling group exhibited a significant reduction in body weight compared to the control group （P<0.05）. After six weeks of treatment， all dosage groups of SQP showed significantly higher body weights compared to the model group （P<0.01）. Compared with the normal group， the model group exhibited significant decreases in bone mineral density （BMD）， bone volume fraction （BV/TV）， trabecular number （Tb.N）， osteocalcin （OCN）， alkaline phosphatase （ALP） levels in femoral head tissue， and serum bone-specific alkaline phosphatase （BALP） （P<0.01）， along with significant increases in trabecular separation （Tb.Sp）， empty lacunae rate in tissue， and apoptosis rate （P<0.01）. In comparison to the model group， the SQP intervention groups showed significant improvements in BMD， BV/TV and Tb.N （P<0.01）， significant reductions in Tb.Sp， empty lacunae rate and apoptosis rate （P<0.05）， and significant increases in protein levels of OCN and ALP as well as BALP content （P<0.05）. The in vitro experimental results revealed that all dosage groups of SQP medicated serum showed no toxic effects on osteoblast. Compared with the normal group， the model group displayed significant suppression of osteoblast proliferation activity， ALP activity， and calcified nodule formation rate （P<0.01）， significant decreases in mRNA transcription levels of OCN and Runt-related transcription factor 2 （RUNX2） （P<0.01）， significant reductions in protein content of osteopontin （OPN）， typeⅠ collagen （ColⅠ）A1， B-cell lymphoma-2 （Bcl-2）， PI3K， and phosphorylated （p）-Akt （P<0.01）， and a significant increase in apoptosis rate （P<0.01）. Compared with the model group， the SQP medicated serum intervention groups exhibited significant increases in proliferation activity， ALP activity， calcified nodule formation rate， mRNA transcription levels of OCN and RUNX2， and protein content of OPN， ColⅠA1， Bcl-2， PI3K， and p-Akt （P<0.05）， along with a significant decrease in apoptosis rate （P<0.01）. ConclusionSQP can effectively reduce the disease severity of SONFH with liver and kidney deficiency syndrome and improve bone microstructure， with the therapeutic effects exhibiting a dose-dependent manner. The mechanism may be related to its regulation of key processes such as muscle and bone metabolism and the correction of imbalances in the "neuro-endocrine-immune" network， thereby promoting osteoblast differentiation and inhibiting osteoblast apoptosis. The PI3K/Akt signaling axis is likely one of the key pathways through which this formula exerts its effects.

Traditional development of small protein scaffolds has relied on display technologies and mutation-based engineering, which limit sequence and functional diversity, thereby constraining their therapeutic and application potential. Protein design tools have significantly advanced the creation of novel protein sequences, structures, and functions. However, further improvements in design strategies are still needed to more efficiently optimize the functional performance of protein-based drugs and enhance their druggability. Here, we extended an evolution-based design protocol to create a novel minibinder, BindHer, against the human epidermal growth factor receptor 2 (HER2). It not only exhibits super stability and binding selectivity but also demonstrates remarkable properties in tissue specificity. Radiolabeling experiments with ^99mTc, ⁶⁸Ga, and ¹⁸F revealed that BindHer efficiently targets tumors in HER2-positive breast cancer mouse models, with minimal nonspecific liver absorption, outperforming scaffolds designed through traditional engineering. These findings highlight a new rational approach to automated protein design, offering significant potential for large-scale applications in therapeutic mini-protein development.

Objective To investigate the relationship between single nucleotide polymorphisms(SNPs)in the eNOS gene and genetic susceptibility to systemic lupus erythematosus(SLE)in Hainan.Methods Blood samples were collected from SLE patients(SLE group,n=214)and healthy controls(control group,n=214)from January 2020 to December 2022 at the First Affiliated Hospital of Hainan Medical College and Hainan Provincial People's Hospital.The bases of eNOS gene rs3918188,rs1799983 and rs1007311 loci in each group were detected by SNaPshot sequencing technology.Logistic regression was used to analyze the correlation between genotypes,alleles and gene models(dominant model,recessive model,and overdominant model)of the above 3 target loci of the eNOS gene and genetic susceptibility to SLE.Haplotype analysis was conducted using HaploView 4.2 software to investigate the relationship between haploid and genetic susceptibility to SLE at each site.Results The results of logistic regression analysis revealed that the CC genotype and the C allele at rs3918188 locus were risk factors for genetic susceptibility to SLE(CC vs.AA:OR=2.449,P<0.05;C vs.A:OR=2.133,P<0.001).In recessive model at rs3918188 locus,CC genotype carriers had an increased risk of SLE development compared with AA+AC genotype carriers(OR=2.774,P<0.001).In contrast,in overdominant model at this locus,AC genotype carriers had a decreased risk of SLE occurrence compared with AA+CC genotype carriers(OR=0.385,P<0.001).In addition,polymorphisms of rs1799983 and rs1007311 were not associated with susceptibility to SLE in genotype,allele type and the 3 genetic models(P>0.05).Haplotype analysis revealed a strong linkage disequilibrium between the rs1007311 and rs1799983 loci of the eNOS gene,but no significant correlation was found between haplotype and genetic susceptibility to SLE(P>0.05).Conclusion The CC genotype and C allele at rs3918188 locus of eNOS gene may be risk factors for SLE in Hainan,while the risk of SLE occurrence is reduced in carriers of AC genotype under the overdominant model.

Objective:To explore the effectiveness of a hysteroscopic endometrial lesion diagnosis model de-veloped based on deep learning(DL)algorithm combined with gradient-weighted class activation mapping(Grad-CAM)visualization technology.Methods:303 hysteroscopy videos(4781 images)of 291 patients who un-derwent hysteroscopy examination in the Department of Gynecology,Renmin Hospital of Wuhan University from June 1,2021 to December 31,2022 were selected.The dataset was divided into a training set(3703 images)and a test set(1078 images)by weight sampling method.After the training set was used for model learning and train-ing,two model architectures,residual neural network(ResNet18)and efficient neural network(EfficientNet-B0),were selected to verify the model in the test set by five-class and two-class classification tasks,respectively.Tak-ing histopathology as the gold standard,the diagnostic efficacy was evaluated to select the optimal model,and the Grad-CAM layer was embedded in the optimal model to output hysteroscopy images of Grad-CAM.Results:①In the five-class classification tasks,the accuracy of EfficientNet-B0 model(93.23％)was higher than that of Res-Net18 model(84.23％);the area under the curve(AUC)of EfficientNet-B0 model in the diagnosis of five disea-ses,including atypical endometrial hyperplasia,endometrial polyps,endometrial cancer,endometrial atypical hy-perplasia,and submucous myoma,was slightly higher than that of ResNet18 model,and the AUC of both models was almost above 0.980.②In the binary classification task of accuracy and the evaluation of specificity,the two models were similar,both above 93.00％,and the sensitivity of EfficientNet-B0 model(91.14％)was significantly better than that of ResNet18 model(77.22％).③EfficientNet-B0 model combined with Grad-CAM algorithm could identify the abnormal areas in the image.After biopsy and pathological examination,it was confirmed that about 95％of the marked areas in the model's output heatmap were lesion areas.Conclusions:The hysteroscopy di-agnostic model developed by EfficientNet-B0 model combined with Grad-CAM has high diagnostic accuracy,sen-sitivity,and specificity,and has application value in the diagnosis of endometrial lesions.

Objective:To explore the effectiveness of a hysteroscopic endometrial lesion diagnosis model de-veloped based on deep learning(DL)algorithm combined with gradient-weighted class activation mapping(Grad-CAM)visualization technology.Methods:303 hysteroscopy videos(4781 images)of 291 patients who un-derwent hysteroscopy examination in the Department of Gynecology,Renmin Hospital of Wuhan University from June 1,2021 to December 31,2022 were selected.The dataset was divided into a training set(3703 images)and a test set(1078 images)by weight sampling method.After the training set was used for model learning and train-ing,two model architectures,residual neural network(ResNet18)and efficient neural network(EfficientNet-B0),were selected to verify the model in the test set by five-class and two-class classification tasks,respectively.Tak-ing histopathology as the gold standard,the diagnostic efficacy was evaluated to select the optimal model,and the Grad-CAM layer was embedded in the optimal model to output hysteroscopy images of Grad-CAM.Results:①In the five-class classification tasks,the accuracy of EfficientNet-B0 model(93.23％)was higher than that of Res-Net18 model(84.23％);the area under the curve(AUC)of EfficientNet-B0 model in the diagnosis of five disea-ses,including atypical endometrial hyperplasia,endometrial polyps,endometrial cancer,endometrial atypical hy-perplasia,and submucous myoma,was slightly higher than that of ResNet18 model,and the AUC of both models was almost above 0.980.②In the binary classification task of accuracy and the evaluation of specificity,the two models were similar,both above 93.00％,and the sensitivity of EfficientNet-B0 model(91.14％)was significantly better than that of ResNet18 model(77.22％).③EfficientNet-B0 model combined with Grad-CAM algorithm could identify the abnormal areas in the image.After biopsy and pathological examination,it was confirmed that about 95％of the marked areas in the model's output heatmap were lesion areas.Conclusions:The hysteroscopy di-agnostic model developed by EfficientNet-B0 model combined with Grad-CAM has high diagnostic accuracy,sen-sitivity,and specificity,and has application value in the diagnosis of endometrial lesions.

Objective:To explore the effectiveness of a hysteroscopic endometrial lesion diagnosis model de-veloped based on deep learning(DL)algorithm combined with gradient-weighted class activation mapping(Grad-CAM)visualization technology.Methods:303 hysteroscopy videos(4781 images)of 291 patients who un-derwent hysteroscopy examination in the Department of Gynecology,Renmin Hospital of Wuhan University from June 1,2021 to December 31,2022 were selected.The dataset was divided into a training set(3703 images)and a test set(1078 images)by weight sampling method.After the training set was used for model learning and train-ing,two model architectures,residual neural network(ResNet18)and efficient neural network(EfficientNet-B0),were selected to verify the model in the test set by five-class and two-class classification tasks,respectively.Tak-ing histopathology as the gold standard,the diagnostic efficacy was evaluated to select the optimal model,and the Grad-CAM layer was embedded in the optimal model to output hysteroscopy images of Grad-CAM.Results:①In the five-class classification tasks,the accuracy of EfficientNet-B0 model(93.23％)was higher than that of Res-Net18 model(84.23％);the area under the curve(AUC)of EfficientNet-B0 model in the diagnosis of five disea-ses,including atypical endometrial hyperplasia,endometrial polyps,endometrial cancer,endometrial atypical hy-perplasia,and submucous myoma,was slightly higher than that of ResNet18 model,and the AUC of both models was almost above 0.980.②In the binary classification task of accuracy and the evaluation of specificity,the two models were similar,both above 93.00％,and the sensitivity of EfficientNet-B0 model(91.14％)was significantly better than that of ResNet18 model(77.22％).③EfficientNet-B0 model combined with Grad-CAM algorithm could identify the abnormal areas in the image.After biopsy and pathological examination,it was confirmed that about 95％of the marked areas in the model's output heatmap were lesion areas.Conclusions:The hysteroscopy di-agnostic model developed by EfficientNet-B0 model combined with Grad-CAM has high diagnostic accuracy,sen-sitivity,and specificity,and has application value in the diagnosis of endometrial lesions.

Objective:To explore the effectiveness of a hysteroscopic endometrial lesion diagnosis model de-veloped based on deep learning(DL)algorithm combined with gradient-weighted class activation mapping(Grad-CAM)visualization technology.Methods:303 hysteroscopy videos(4781 images)of 291 patients who un-derwent hysteroscopy examination in the Department of Gynecology,Renmin Hospital of Wuhan University from June 1,2021 to December 31,2022 were selected.The dataset was divided into a training set(3703 images)and a test set(1078 images)by weight sampling method.After the training set was used for model learning and train-ing,two model architectures,residual neural network(ResNet18)and efficient neural network(EfficientNet-B0),were selected to verify the model in the test set by five-class and two-class classification tasks,respectively.Tak-ing histopathology as the gold standard,the diagnostic efficacy was evaluated to select the optimal model,and the Grad-CAM layer was embedded in the optimal model to output hysteroscopy images of Grad-CAM.Results:①In the five-class classification tasks,the accuracy of EfficientNet-B0 model(93.23％)was higher than that of Res-Net18 model(84.23％);the area under the curve(AUC)of EfficientNet-B0 model in the diagnosis of five disea-ses,including atypical endometrial hyperplasia,endometrial polyps,endometrial cancer,endometrial atypical hy-perplasia,and submucous myoma,was slightly higher than that of ResNet18 model,and the AUC of both models was almost above 0.980.②In the binary classification task of accuracy and the evaluation of specificity,the two models were similar,both above 93.00％,and the sensitivity of EfficientNet-B0 model(91.14％)was significantly better than that of ResNet18 model(77.22％).③EfficientNet-B0 model combined with Grad-CAM algorithm could identify the abnormal areas in the image.After biopsy and pathological examination,it was confirmed that about 95％of the marked areas in the model's output heatmap were lesion areas.Conclusions:The hysteroscopy di-agnostic model developed by EfficientNet-B0 model combined with Grad-CAM has high diagnostic accuracy,sen-sitivity,and specificity,and has application value in the diagnosis of endometrial lesions.

Objective:To explore the effectiveness of a hysteroscopic endometrial lesion diagnosis model de-veloped based on deep learning(DL)algorithm combined with gradient-weighted class activation mapping(Grad-CAM)visualization technology.Methods:303 hysteroscopy videos(4781 images)of 291 patients who un-derwent hysteroscopy examination in the Department of Gynecology,Renmin Hospital of Wuhan University from June 1,2021 to December 31,2022 were selected.The dataset was divided into a training set(3703 images)and a test set(1078 images)by weight sampling method.After the training set was used for model learning and train-ing,two model architectures,residual neural network(ResNet18)and efficient neural network(EfficientNet-B0),were selected to verify the model in the test set by five-class and two-class classification tasks,respectively.Tak-ing histopathology as the gold standard,the diagnostic efficacy was evaluated to select the optimal model,and the Grad-CAM layer was embedded in the optimal model to output hysteroscopy images of Grad-CAM.Results:①In the five-class classification tasks,the accuracy of EfficientNet-B0 model(93.23％)was higher than that of Res-Net18 model(84.23％);the area under the curve(AUC)of EfficientNet-B0 model in the diagnosis of five disea-ses,including atypical endometrial hyperplasia,endometrial polyps,endometrial cancer,endometrial atypical hy-perplasia,and submucous myoma,was slightly higher than that of ResNet18 model,and the AUC of both models was almost above 0.980.②In the binary classification task of accuracy and the evaluation of specificity,the two models were similar,both above 93.00％,and the sensitivity of EfficientNet-B0 model(91.14％)was significantly better than that of ResNet18 model(77.22％).③EfficientNet-B0 model combined with Grad-CAM algorithm could identify the abnormal areas in the image.After biopsy and pathological examination,it was confirmed that about 95％of the marked areas in the model's output heatmap were lesion areas.Conclusions:The hysteroscopy di-agnostic model developed by EfficientNet-B0 model combined with Grad-CAM has high diagnostic accuracy,sen-sitivity,and specificity,and has application value in the diagnosis of endometrial lesions.

Objective:To explore the effectiveness of a hysteroscopic endometrial lesion diagnosis model de-veloped based on deep learning(DL)algorithm combined with gradient-weighted class activation mapping(Grad-CAM)visualization technology.Methods:303 hysteroscopy videos(4781 images)of 291 patients who un-derwent hysteroscopy examination in the Department of Gynecology,Renmin Hospital of Wuhan University from June 1,2021 to December 31,2022 were selected.The dataset was divided into a training set(3703 images)and a test set(1078 images)by weight sampling method.After the training set was used for model learning and train-ing,two model architectures,residual neural network(ResNet18)and efficient neural network(EfficientNet-B0),were selected to verify the model in the test set by five-class and two-class classification tasks,respectively.Tak-ing histopathology as the gold standard,the diagnostic efficacy was evaluated to select the optimal model,and the Grad-CAM layer was embedded in the optimal model to output hysteroscopy images of Grad-CAM.Results:①In the five-class classification tasks,the accuracy of EfficientNet-B0 model(93.23％)was higher than that of Res-Net18 model(84.23％);the area under the curve(AUC)of EfficientNet-B0 model in the diagnosis of five disea-ses,including atypical endometrial hyperplasia,endometrial polyps,endometrial cancer,endometrial atypical hy-perplasia,and submucous myoma,was slightly higher than that of ResNet18 model,and the AUC of both models was almost above 0.980.②In the binary classification task of accuracy and the evaluation of specificity,the two models were similar,both above 93.00％,and the sensitivity of EfficientNet-B0 model(91.14％)was significantly better than that of ResNet18 model(77.22％).③EfficientNet-B0 model combined with Grad-CAM algorithm could identify the abnormal areas in the image.After biopsy and pathological examination,it was confirmed that about 95％of the marked areas in the model's output heatmap were lesion areas.Conclusions:The hysteroscopy di-agnostic model developed by EfficientNet-B0 model combined with Grad-CAM has high diagnostic accuracy,sen-sitivity,and specificity,and has application value in the diagnosis of endometrial lesions.

Objective:To explore the effectiveness of a hysteroscopic endometrial lesion diagnosis model de-veloped based on deep learning(DL)algorithm combined with gradient-weighted class activation mapping(Grad-CAM)visualization technology.Methods:303 hysteroscopy videos(4781 images)of 291 patients who un-derwent hysteroscopy examination in the Department of Gynecology,Renmin Hospital of Wuhan University from June 1,2021 to December 31,2022 were selected.The dataset was divided into a training set(3703 images)and a test set(1078 images)by weight sampling method.After the training set was used for model learning and train-ing,two model architectures,residual neural network(ResNet18)and efficient neural network(EfficientNet-B0),were selected to verify the model in the test set by five-class and two-class classification tasks,respectively.Tak-ing histopathology as the gold standard,the diagnostic efficacy was evaluated to select the optimal model,and the Grad-CAM layer was embedded in the optimal model to output hysteroscopy images of Grad-CAM.Results:①In the five-class classification tasks,the accuracy of EfficientNet-B0 model(93.23％)was higher than that of Res-Net18 model(84.23％);the area under the curve(AUC)of EfficientNet-B0 model in the diagnosis of five disea-ses,including atypical endometrial hyperplasia,endometrial polyps,endometrial cancer,endometrial atypical hy-perplasia,and submucous myoma,was slightly higher than that of ResNet18 model,and the AUC of both models was almost above 0.980.②In the binary classification task of accuracy and the evaluation of specificity,the two models were similar,both above 93.00％,and the sensitivity of EfficientNet-B0 model(91.14％)was significantly better than that of ResNet18 model(77.22％).③EfficientNet-B0 model combined with Grad-CAM algorithm could identify the abnormal areas in the image.After biopsy and pathological examination,it was confirmed that about 95％of the marked areas in the model's output heatmap were lesion areas.Conclusions:The hysteroscopy di-agnostic model developed by EfficientNet-B0 model combined with Grad-CAM has high diagnostic accuracy,sen-sitivity,and specificity,and has application value in the diagnosis of endometrial lesions.