This case report presents a 16-year-old male patient with deficiency of adenosine deaminase 2(DADA2). The patient had a history of Raynaud′s phenomenon with digital ulcers since childhood. As the disease progressed, the patient developed retinal vasculitis, intracranial hemorrhage, skin necrosis, severe malnutrition, refractory hypertension, and gastrointestinal bleeding. Genetic testing revealed compound heterozygous mutations in the ADA2 gene (paternal c.1072G > A pathogenic mutation + maternal c.1065C > A variant of unknown clinical significance). ADA2 enzyme activity was significantly reduced (0.1 U/L). A multidisciplinary treatment team developed an individualized plan to address key issues, including nutritional support, blood pressure control, rehabilitation, pain management, and balancing anticoagulation/bleeding risks. After systematic intervention, the patient′s condition showed partial improvement. This case reveals clinical challenges such as anticoagulation decisions and nutritional support of DADA2. It also emphasizes the importance of early molecular diagnosis, tumor necrosis factor-α inhibitor targeted therapy, and multidisciplinary collaboration in management, underlining the core value of precision medicine in rare disease management.

Objective To investigate the liver disease phenotypes and clinical features of patients with different genotypes of Wilson's disease(WD).Methods A retrospective analysis was performed for 163 patients with WD who were diagnosed and underwent genetic testing in The Fifth Medical Center of Chinese PLA General Hospital from August 2008 to June 2023,and clinical manifestations,laboratory examination,pathological examination,imaging examination,and ATP7B genetic testing results were collected.According to ATP7B gene mutation,the patients were divided into groups as follows:R778L mutation group and non-R778L mutation group;P992L mutation group and non-P992L mutation group;truncation mutation group and non-truncation mutation group.Liver disease phenotypes and clinical features were analyzed for the patients with c.2333G>T/p.R778L mutation(R778L mutation),c.2975C>T/p.P992L mutation(P992L mutation),and truncation mutation of the ATP7B gene.The Mann-Whitney U test or the Kruskal-Wallis H test was used for comparison of continuous data between groups,and the chi-square test or the Fisher's exact test was used for comparison of categorical data between groups.Results The 163 patients with WD had varying severities of liver disease phenotypes,among whom 121(74.23%)were diagnosed with chronic liver disease,36(22.09%)were diagnosed with decompensated cirrhosis,and 6(3.68%)were diagnosed with fulminant WD,and in addition,there were 5 patients(2 with chronic liver disease and 3 with decompensated cirrhosis)with neurological abnormalities.For the 163 patients with WD,R778L mutation(with an allele frequency of 28.2%)was the most common mutation in the ATP7B gene,followed by P992L mutation(with an allele frequency of 12.6%),and truncation mutation showed an allele frequency of 11.0%.There was no significant difference in the distribution of the three mutations across different liver disease phenotypes(P>0.05).The R778L mutation group had a significantly lower level of ceruloplasmin(CP)than the non-R778L mutation group[0.04(0.02-0.08)g/L vs 0.08(0.03-0.13)g/L,Z=-2.889,P=0.004].Compared with the non-P992L mutation group,the P992L mutation group had significantly higher levels of alanine aminotransferase[135.0(80.5-237.0)U/L vs 80.5(36.0-173.3)U/L,Z=2.684,P=0.007]and aspartate aminotransferase[121.4(77.0-195.0)U/L vs 84.0(39.0-123.3)U/L,Z=3.388,P<0.001].Compared with the non-truncation mutation group,the truncation mutation group had significantly lower levels of CP[0.03(0.02-0.08)g/L vs 0.06(0.03-0.11)g/L,Z=-3.136,P=0.002]and serum copper[3.20(2.15-5.00)mg/L vs 4.20(2.60-7.50)mg/L,Z=-2.296,P=0.025].Conclusion R778L mutation,P992L mutation and truncation mutation are not associated with liver disease phenotype in WD patients;however,R778L mutation is associated with a lower level of CP,P992L mutation is associated with higher levels of ALT and AST,and truncation mutation is associated with lower levels of CP and serum copper.

Objective:To investigate the relationship between intramuscular adipose tissue index (IATI) calculated from computed tomography images at transverse process of the first lumbar and all-cause mortality in maintenance dialysis patients, and to provide a reference for improving the prognosis in these patients.Methods:It was a multicenter retrospective cohort study. The clinical data of patients who received maintenance hemodialysis or peritoneal dialysis treatment from January 1, 2017 to December 31, 2019 in 4 grade Ⅲ hospitals including Zhongda Hospital Affiliated to Southeast University, Taizhou People's Hospital Affiliated to Nanjing Medical University, Affiliated Hospital of Yangzhou University, and the Third Affiliated Hospital of Soochow University were retrospectively collected. IATI was calculated by low attenuation muscle (LAM) density/skeletal muscle density. The receiver-operating characteristic curve was used to determine the optimal cut-off value of IATI, and the patients were divided into high IATI group and low IATI group according to the optimal cut-off value. The differences of baseline clinical data and measurement parameters of the first lumbar level between the two groups were compared. The follow-up ended on December 23, 2022. The endpoint event was defined as all-cause mortality within 3 years. Kaplan-Meier survival curve and log-rank test were used to analyze the survival rates and the differences between the two groups. Multivariate Cox regression analysis models were used to analyze the association between IATI and the risk of all-cause mortality in maintenance dialysis patients. Multivariate logistic regression analysis model was used to analyze the influencing factors of high IATI.Results:A total of 478 patients were eligibly recruited in this study, with age of (53.55±13.19) years old and 319 (66.7%) males, including 365 (76.4%) hemodialysis patients and 113 (23.6%) peritoneal dialysis patients. There were 376 (78.7%) patients in low IATI (<0.42) group and 102 (21.3%) patients in high IATI (≥0.42) group. The proportion of age ≥ 60 years old ( χ2=24.746, P<0.001), proportion of diabetes mellitus ( χ2=5.570, P=0.018), fasting blood glucose ( t=-2.145, P=0.032), LAM density ( t=-3.735, P<0.001), LAM index ( t=-7.072, P<0.001), and LAM area/skeletal muscle area ratio ( Z=-9.630, P<0.001) in high IATI group were all higher than those in low IATI group, while proportion of males ( χ2=11.116, P<0.001), serum albumin ( Z=2.708, P=0.007) and skeletal muscle density ( t=12.380, P<0.001) were lower than those in low IATI group. Kaplan-Meier survival analysis showed that the 3-years overall survival rate of low IATI group was significantly higher than that in high IATI group (Log-rank χ2=19.188, P<0.001). Multivariate Cox regression analysis showed that IATI<0.42 [<0.42/≥0.42, HR(95% CI): 0.50 (0.31-0.83), P=0.007] was an independent protective factor of all-cause mortality, and age ≥60 years old [ HR (95% CI): 2.61 (1.60-4.23), P<0.001], diabetes mellitus [ HR (95% CI): 1.71 (1.06-2.78), P=0.029] and high blood neutrophil/lymphocyte ratio [ HR (95% CI): 1.04 (1.00-1.07), P=0.049] were the independent risk factors of all-cause mortality in maintenance dialysis patients. Stepwise Cox regression analysis showed that IATI<0.42 was still an independent protective factor of all-cause mortality in maintenance dialysis patients [<0.42/≥0.42, HR (95% CI): 0.45 (0.27-0.76), P=0.003]. Multivariate logistic regression analysis showed that low skeletal muscle density [ OR (95% CI): 0.84 (0.81-0.88), P<0.001] and high serum triglyceride [ OR (95% CI): 1.39 (1.07-1.82), P=0.015] were the independent influencing factors of IATI≥0.42. Conclusion:IATI<0.42 of the first lumbar level is an independent protective factor of all-cause mortality in maintenance dialysis patients. Localized myosteatosis within high-quality skeletal muscle may reduce the risk of all-cause mortality in these patients.

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.