Objective:To investigate the characteristics of weight gain in the mid and late pregnancy of women with early pregnancy weight loss, and the relationship between weight gain and weight gain rate before the diagnosis of gestational diabetes mellitus (GDM) and GDM.Methods:A retrospective study was conducted on 2 614 singleton pregnant women who underwent prenatal care and delivered at Beijing Obstetrics and Gynecology Hospital, Capital Medical University, from January 2014 to December 2021, and whose weight decreased compared to pre-pregnancy weight by 16 weeks of gestation. The rate of weight gain in mid and late pregnancy, also weight gain and weight gain rate at each stage were analyzed. Multivariate logistic regression was used to analyze the impact of total weight gain and weight gain rate at each stage before the diagnosis of GDM on the risk of GDM.Results:The rates of weight gain in mid and late pregnancy for women with early pregnancy weight loss who were underweight, normal weight, overweight, and obese before pregnancy were (0.60±0.15), (0.59±0.18), (0.53±0.20), and (0.42±0.20) kg/week, respectively, all higher than the "Chinese Recommended Standards for Weight Gain During Pregnancy" [which are (0.37-0.56), (0.26-0.48), (0.22-0.37), and (0.15-0.30) kg/week, respectively]. The weight gain rates at each stage of mid and late pregnancy in women with early pregnancy weight loss showed a "bimodal" trend, with the first peak in weight gain rate occurring at 16-20 or 20-24 weeks of gestation [weight gain rates for underweight, normal weight, overweight, and obese women before pregnancy were 0.75 kg/week (0.44-1.00 kg/week), 0.74 kg/week (0.50-1.00 kg/week), 0.63 kg/week (0.39-0.86 kg/week), and 0.50 kg/week (0.25-0.74 kg/week), respectively] and the second peak occurring at 28-32 weeks [weight gain rates for underweight, normal weight, overweight, and obese women before pregnancy were 0.63 kg/week (0.50-1.00 kg/week), 0.63 kg/week (0.38- 0.88 kg/week), 0.60 kg/week (0.25-0.88 kg/week), and 0.50 kg/week (0.22-0.75 kg/week). As of 28 weeks and 36 weeks of gestation, 53.7% (1 404/2 614) and 77.4% (1 946/2 512) of pregnant women, respectively, reached the lower limit of the recommended weight gain for the corresponding gestational weeks. No association was found between insufficient weight gain ( aOR=0.828, 95% CI: 0.639-1.071, P=0.151) or excessive weight gain ( aOR=0.936, 95% CI: 0.598-1.465, P=0.773) before the diagnosis of GDM and the risk of GDM. However, obese women with a weight loss greater than 5% of their pre-pregnancy weight in early pregnancy and a rapid weight gain rate (> P 75) between 16-20 weeks of gestation had an increased risk of developing GDM ( aOR=32.870, 95% CI: 1.625-664.775, P=0.023). Conclusions:In clinical practice, dynamic monitoring of weight changes at various stages of pregnancy in women who lose weight in early pregnancy is recommended. Targeted weight management during mid-pregnancy for women who are obese before pregnancy and experience significant weight loss in early pregnancy may help prevent excessive gestational weight gain and decrease the risk of GDM.

Objective:To describe the clinical manifestations, genetic mutation site, diagnosis, and treatment of a patient with multisystem proteinopathy type 1 (MSP1) caused by valosin-containing protein ( VCP) gene mutation, and to improve clinicians′ understanding of this disease. Methods:A retrospective analysis was performed on clinical and genetic data from a confirmed VCP gene missense mutation-associated MSP1 case diagnosed at the Department of Neurology, Affiliated Zhongda Hospital, School of Medicine, Southeast University in January 2024. A 12-month follow-up and systematic literature review were performed for comprehensive analysis. Results:The 53-year-old male patient presented with progressive limb weakness over 7 months. Neurological examination demonstrated tongue fasciculations, asymmetric proximal muscle weakness in all four limbs, left patellar hyperreflexia, positive right Chaddock sign, and bilateral Hoffmann signs. Electrophysiological studies demonstrated extensive neurogenic damage. Lower-limb muscle magnetic resonance imaging (MRI) showed selective fatty infiltration in specific muscle groups. Biceps brachii biopsy pathology revealed rimmed vacuoles and grouped atrophy of typeⅡfibers. Immunofluorescence confirmed aberrant aggregation of VCP within atrophic myofibers, showing co-localization with p62 and transactive response DNA binding protein 43 (TDP-43). Whole-genome sequencing identified a heterozygous c.463C>T (p.Arg155Cys) missense mutation in exon 5 of the VCP gene, classified as a likely pathogenic mutation according to the guidelines of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. The patient was diagnosed with MSP1 with amyotrophic lateral sclerosis and inclusion body myopathy as the main clinical manifestation based on clinical manifestations, electrophysiology, imaging, histopathology, and genetic findings. After 12 months of riluzole therapy, disease progression remained relatively slow. Literature review identified 67 relevant articles, revealing 87 VCP mutation genotypes and 19 clinical phenotypes. Conclusions:MSP1 is a genetically and phenotypically heterogeneous spectrum of multisystem degenerative disorders. This case represents the first reported VCP-related MSP1 in China, characterized by amyotrophic lateral sclerosis combined with inclusion body myopathy. Riluzole treatment demonstrates slowed disease progression over 1 year.

Objective To investigate the value of machine learning models based on contrast-enhanced CT radiomics in differentia-ting renal oncocytoma(RO)from chromophobe renal cell carcinoma(chRCC).Methods A total of 65 patients with RO and chRCC confirmed by pathology with complete clinical and imaging data were analyzed retrospectively.The patients were randomly divided into training set(n=45)and test set(n=20)according to a ratio of 7︰3.The tumor boundaries were delineated on the preoperative CT images using 3D Slicer software,and radiomics features were extracted using the Radiomics plugin.Univariate analysis,recursive fea-ture elimination(RFE),least absolute shrinkage and selection operator(LASSO)algorithms were used to select the best radiomics features.Three machine learning models were constructed on the training set and the grid search method was used to select the best combination of hyperparameters.The receiver operating characteristic(ROC)curve,calibration curve and decision curve were used to evaluate the performance of each machine learning model on the training set and test set.Results Random forest model,logistic regres-sion model and support vector machine model can better identify RO and chRCC.In the training set,the area under the curve(AUC)of random forest model and support vector machine model were 0.950[95%confidence interval(CI)0.901-0.998]and 0.955(95%CI 0.908-1.000),respectively,which were higher than the AUC of logistic regression model 0.882(95%CI 0.806-0.956).Statistical differences were found by DeLong test(P<0.05);In the test set,the AUC of random forest model,logistic regression model and support vector machine model were 0.876(95%CI 0.758-0.993),0.883(95%CI 0.768-0.997)and 0.883(95%CI 0.768-0.997),respectively.There was no significant statistical difference in the AUC of each model by DeLong test(P>0.05).The decision curve showed that all three models had significant net clinical benefits.Conclusion The machine learning model based on contrast-enhanced CT radiomics can effectively distinguish RO from chRCC.

Objective:To investigate the relationship between the early pregnancy triglyceride-glucose (TyG) index and gestational diabetes mellitus (GDM) in twin pregnancies.Methods:This retrospective study involved twin-pregnant women who visited Beijing Obstetrics and Gynecology Hospital, Capital Medical University from October 2015 to February 2021. Based on the results of the 75 g oral glucose tolerance test (OGTT) performed at 24-28 weeks of gestation, the women were divided into the GDM and the control groups. The groups were further stratified based on maternal age (<35 years or ≥35 years), pre-pregnancy body mass index (BMI) (<24.0 or ≥24.0 kg/m2), and conception method [assisted reproductive technology (ART) or natural conception]. The correlation between early pregnancy TyG index and GDM, as well as the predictive value of the early pregnancy TyG index for the risk of GDM in twin pregnancies, were analyzed. The TyG index in early pregnancy was then divided into tertiles, and the risks of GDM in low, medium, and high TyG index groups were analyzed. Statistical analyses were performed using independent sample t-test, non-parametric test, Chi-square test, and binary logistic regression analysis. Receiver operating characteristic (ROC) curve analysis was used to evaluate the predictive value of the early pregnancy TyG index for GDM in twin pregnancies. Results:(1) A total of 1 684 twin-pregnant women were included, with an average age of 32.3 years (29.8-34.9 years) and a pre-pregnancy BMI of 22.0 kg/m2 (20.0-24.3 kg/m2). Among them, 319 (18.9%) were multiparas, 982 (58.3%) conceived through ART, and 357 (21.2%) were monochorionic twins. Of the 1 684 women, 367 (21.8%) were diagnosed with GDM (GDM group), whereas the remaining 1 317 were classified as the control group. (2) Compared to the control group, the GDM group had older maternal age [(32.2±3.7) years vs. (33.3±3.8) years, t=－4.92], higher pre-pregnancy weight, and BMI [57.5 kg (52.0-65.0 kg) vs. 60.0 kg (55.0-67.3 kg), U=279 901.50; 21.8 kg/m2 (19.8-24.0 kg/m2) vs. 22.9 kg/m2 (20.9-25.5 kg/m2), U=288 435.00]. The proportions of a family history of diabetes, history of GDM and polycystic ovary syndrome (PCOS) were all higher in the GDM group compared to the control group [9.6% (127/1 317) vs. 19.1% (70/367), χ 2=24.71; 0.8% (2/1 317) vs. 10.8% (8/367), χ 2=20.00; 9.1% (120/1 317) vs. 15.3% (56/367), χ 2=11.59] (all P<0.001). The GDM group had higher early pregnancy fasting blood glucose, triglyceride, and TyG indices compared to the control group [4.51 mmol/L (4.28-4.75 mmol/L) vs. 4.68 mmol/L (4.42-4.97 mmol/L), U=7.14; 1.23 mmol/L (0.93-1.57 mmol/L) vs. 1.43 mmol/L (1.09-1.89 mmol/L), U=4.81; 8.39±0.41 vs. 8.59±0.43, t=6.46]. The incidence of gestational anemia and weight gain were lower in the GDM group compared to the control group [39.2% (516/1 317) vs. 33.0% (121/367), χ 2=4.71; 17.0 kg (13.7-20.5 kg) vs. 15.0 kg (12.0-18.3 kg), U=187 966.00] (all P<0.05). The proportion of male newborns in the GDM group was higher than in the control group [52.5% (1 384/2 634) vs. 46.7% (343/734), χ 2=7.77, P=0.005]. (3) Early pregnancy TyG index was associated with GDM in twin pregnancies ( OR=3.164, 95% CI: 2.371-4.220, P<0.001). After adjusting for maternal age, pre-pregnancy BMI, history of GDM, history of macrosomia, and family history of diabetes, the early pregnancy TyG index remained associated with GDM ( OR=2.560, 95% CI: 1.884-3.478, P<0.001). Analysis of the early pregnancy TyG index divided into tertiles (corresponding TyG indices of 8.25 and 8.59) revealed that, compared to those with a low TyG index, those with a mid TyG index had a 0.555-fold increased risk of GDM ( OR=1.555, 95% CI: 1.119-2.159, P=0.008), and those with a high TyG index had a 1.564-fold increased risk of GDM ( OR=2.564, 95% CI: 1.836-3.530, P<0.001). Stratified analysis by age, BMI, and mode of conception showed that the early pregnancy TyG index was associated with GDM in twin pregnancies (all P<0.001). (4) The threshold value for the early pregnancy TyG index to predict GDM in twin pregnancies was 8.33, with an area under the curve (AUC) of 0.632, 95% CI: 0.600-0.665, sensitivity of 0.744, and specificity of 0.436. The AUC in twin pregnancies for those who conceived via ART was 0.635 (95% CI: 0.593-0.676, P<0.001), slightly higher than in those who conceived naturally (AUC=0.628, 95% CI: 0.576-0.681, P<0.001). After adjusting for maternal age, pre-pregnancy BMI, history of GDM, and family history of diabetes, the AUC for the early pregnancy TyG index to predict GDM in twin pregnancies was 0.675 (95% CI: 0.644-0.707). For those who conceived via ART, the AUC (95% CI) was 0.675 (0.634-0.717), slightly lower than for those who conceived naturally [0.682 (0.632-0.733)] (all P<0.001). Conclusion:A high TyG index in the first trimester is a risk factor for GDM in twin pregnancies, but its predictive value for GDM in twin pregnancies needs further research to be confirmed.

OBJECTIVE: To investigate the impact of early antimicrobial therapy on the prognosis of patients with suspected sepsis in emergency and outpatient settings. METHODS: A prospective cohort study was conducted. Patients with suspected sepsis admitted to the emergency department of Taizhou Hospital, Zhejiang Province, from May 1, 2022, to July 31, 2023, were enrolled. Participants were divided into an early group (0-1 hour) and a delayed group (> 1 hour) according to duration from admission to antimicrobial administration. General information, initial vital signs, laboratory parameters within 24 hours after admission, disease severity scores, vasoactive drug usage, and clinical outcomes of the patient were collected. Kaplan-Meier survival curve was used to analyze 28-day survival. Multivariate Cox proportional hazards regression was performed to identify independent risk factors for prognosis of the patients with suspected sepsis in emergency and outpatient settings. Sensitivity analyses were conducted through subgroup analyses. RESULTS: A total of 143 patients with suspected sepsis were enrolled in the analysis, with 66 patients in the early group and 77 in the delayed group. No statistically significant differences were observed in baseline characteristics (age, gender, vital signs, laboratory parameters, disease severity scores) or clinical outcomes [vasoactive drug usage rate, mechanical ventilation duration, length of intensive care unit (ICU) stay, total hospitalization duration] between the two groups. The 28-day mortality, multidrug resistance rate and sepsis confirmation rate did not differ significantly between the early group and delay group [28-day mortality: 18.2% (12/66) vs. 20.8% (16/77), multidrug resistance rate: 3.0% (2/66) vs. 2.6% (2/77), sepsis confirmation rate: 87.9% (58/66) vs. 88.3% (68/77), all P > 0.05]. Kaplan-Meier survival curve analysis showed no difference in 28-day cumulative survival between the two groups (Log-Rank test: χ² = 2.528, P = 0.112). Multivariate Cox proportional hazards regression identified vasoactive drug usage [hazard ration (HR) = 2.465, 95% confidence interval (95%CI) was 1.019-5.961, P = 0.045] and endotracheal intubation (HR = 5.516, 95%CI was 2.195-13.858, P < 0.001) as independent risk factors for 28-day death of the patients with suspected sepsis in emergency and outpatient settings. Further exploration of the impact of early antimicrobial therapy on 28-day death in different subgroups of the patients with suspected sepsis in emergency and outpatient settings was conducted through subgroup analysis. The results showed that in the patients with different ages (< 60 years old: HR = 1.214, 95%CI was 0.535-2.751, P = 0.643; ≥ 60 years old: HR = 2.085, 95%CI was 0.233-18.668, P = 0.511), sequential organ failure assessment (SOFA) scores (< 6: HR = 1.411, 95%CI was 0.482-4.128, P = 0.530; ≥ 6: HR = 0.869, 95%CI was 0.292-2.587, P = 0.801), shock indexes (< 1: HR = 1.095, 95%CI was 0.390-3.077, P = 0.863; ≥ 1: HR = 1.364, 95%CI was 0.458-4.059, P = 0.577) and whether diagnosed with sepsis or not (yes: HR = 0.943, 95%CI was 0.059-15.091, P = 0.967; no: HR = 1.207, 95%CI was 0.554-2.628, P = 0.636) subgroups, early usage of antibiotics had not shown any advantage in improving prognosis compared with delayed usage. CONCLUSION Early antimicrobial therapy does not improve the prognosis of patients with suspected sepsis in emergency and outpatient settings.
Humans ; Sepsis/drug therapy* ; Prospective Studies ; Prognosis ; Emergency Service, Hospital ; Outpatients ; Female ; Male ; Anti-Infective Agents/therapeutic use* ; Middle Aged ; Aged ; Proportional Hazards Models ; Treatment Outcome

Objective: To explore the network structure of dietary intake and physical fitness subtypes among junior and senior high school students, so as to provide precise and efficient intervention guidance for improving abnormal physical health issues in adolescents. Methods: Based on the 2020-2021 Database of Youth Health (DYH), a total of 9 730 junior and senior high school students in Shandong Province were included for analysis. Dietary intake was assessed using a modified Chinese Dietary Quality Questionnaire, and physical fitness was evaluated according to the "2014 National Student Physical Fitness Standard". The Kmeans clustering algorithm was employed to identify potential subtypes of physical fitness in junior and high school students. Network analysis was used to construct a network linking dietary intake and physical fitness symptoms and to identify clusters of dietary behaviors and physical fitness symptoms among adolescents of different genders. Results: The physical fitness among junior and senior high school students of different genders were categorized into two subtypes: the baseline steady type (3 427 male students, 4 039 female students) and the morphological deviation type (1 294 male students, 970 female students). Statistically significant differences were observed in network strength and network connectivity strength among different genders and physical fitness subtypes (all P<0.05). Connections between food intake and physical health symptom clusters across different genders and physical health types among junior and senior high school students were primarily achieved through instant noodle intake and physical fitness (males of morphological deviation type, weight=0.06), fruit intake and physical fitness (males of baseline steady type, weight=-0.07), potato intake and vital capacity (females of morphological deviation type, weight=0.09), and processed meat intake and vital capacity (females of baseline steady type, weight=0.05). Conclusions Dietary intake serves as a significant modifiable risk factor for the physical fitness of junior and high school students. Interventions should focus on promoting healthy eating habits.

Objective:This study aimed to evaluate the effects of hypertensive disorders of pregnancy (HDP), including their clinical subtypes, on neonatal heel blood methionine levels and explore potential dose-effect relationships.Methods:A retrospective cohort study was conducted among 11 007 singleton pregnancies and their neonates delivered at Beijing Obstetrics and Gynecology Hospital, Capital Medical University, from July 2021 to October 2022. Participants were stratified into an HDP group [ n=992; 480 with gestational hypertension, 512 with preeclampsia (including 229 severe cases)] and a non-HDP control group ( n=10 015). Methionine concentrations were measured using tandem mass spectrometry from heel blood dried filter paper samples collected within 72 hours post-delivery. Statistical analyses included non-parametric tests to compare intergroup differences, multiple linear regression to evaluate the effects of HDP on methionine levels, and multivariate logistic regression to identify risk factors for hypermethioninemia (>50 μmol/L). Results:(1) Baseline data: Maternal age was higher in the HDP group compared to controls [33 (30-36) vs. 33 (30-35) years, Z=－2.29, P=0.022], with elevated pre-pregnancy body mass index (BMI) [23 (21-26) vs. 21 (20-23) kg/m2, Z=－17.15, P<0.001] and increased gestational hyperglycemia prevalence [26.5% (263/992) vs. 19.8% (1 986/10 015), χ2=27.95, P<0.001]. (2) Methionine level: Neonates in the HDP group exhibited higher methionine levels [25.96 (21.58-30.89) vs. 24.77 (20.45-29.53) μmol/L, Z=－5.26, P<0.001], with a severity-dependent gradient: gestational hypertension [25.83 (21.77-30.61)], preeclampsia [26.05 (21.23-31.11)], and severe preeclampsia [26.15 (21.25-32.13)] ( Z=2.97, 3.92, 2.26; all P<0.05). Trend analysis confirmed a dose-effect relationship between HDP and neonatal methionine ( χ2=7.82, P=0.005). (3) Multivariate analysis: After adjusting for confounding factors such as maternal age and BMI, HDP remained independently associated with elevated methionine levels ( β=0.93, 95% CI: 0.47-1.40, t=3.92, P<0.001) and increased hypermethioninemia risk ( OR=2.75, 95% CI: 1.13-6.68). Subgroup analysis revealed ORs of 3.20 (95% CI: 1.07-9.57) for gestational hypertension, 3.25 (95% CI: 1.09-9.72) for preeclampsia, and 5.23 (95% CI: 1.54-17.82) for severe preeclampsia (all P<0.05). (4) Neonatal outcomes: Neonates in the HDP group had lower birth weights [3 230 (2 910-3 560) vs. 3 335 (3 070-3 600) g, Z=－7.43, P<0.001] and higher fetal growth restriction rates [10.3% (102/992) vs. 3.1% (306/10 015), χ2=136.47, P<0.001]. Conclusions:HDP demonstrates an elevation of neonatal methionine levels, correlating with disease severity, particularly in severe preeclampsia. These findings underscore the necessity for enhanced metabolic monitoring and long-term follow-up in offspring of mothers with HDP, especially those with severe preeclampsia.

Objective To predict the mechanism of herbal couple Curcumae Rhizoma-Sparganii Rhizoma(CR-SR)in modulation of angiogenesis against lung cancer based on network pharmacology and molecular docking technology,and validate by zebrafish model.Methods The active ingredients and potential targets for anti-lung cancer and antiangiogenesis of CR and SR were screened by network pharmacology.The targets were intersected with those screened from the OMIM database and GeneCards database for lung cancer and antiangiogenesis.Herbal couple-lung cancer and herbal couple-antiangiogenesis of protein-protein interaction(PPI)network was constructed by taking intersecting targets to screen the common and core targets of the herbal couple in lung cancer and anti-angiogenesis.Herbal couple-lung cancer and herbal couple-antiangiogenesis of Gene ontology(GO)function and Kyoto encyclopedia of genes and genomes(KEGG)pathway enrichment analyses were performed by Metascape database.The binding ability and the amino acid residues involved of core targets to major components were evaluated by molecular docking technique.In vitro,CCK-8 method was applied to investigate the effects of herbal couple and single drugs on the cell viability of human umbilical vein endothelial cells(HUVECs).Zebrafish embryos were randomly divided into blank control group,different concentration of drug pairs and single drug group,and positive drug control group,and the number of intersegmental vessels of zebrafish in each group was counted after 72 hour.The mRNA expression levels of angiogenesis-related genes,VEGFA,VEGFR2,VEGFR3,EGFR,etc.,were detected by qRT-PCR.Results 106 herbal couple-lung cancer common targets and 130 herbal couple-antiangiogenesis common targets were screened by network pharmacology.Meanwhile,85 of targets were identical.GO function enrichment analyses of herbal couple-lung cancer resulted in 1648 GO analysis entries,KEGG pathway enrichment analyses resulted in 186 signaling pathways.GO function enrichment analyses herbal couple-antiangiogenesis resulted in 1844 GO analysis entries,KEGG pathway enrichment analyses resulted in 188 signaling pathways.The molecular docking results showed a better affinity between the target and the components,and the forces between them mainly included hydrogen bonding and hydrophobic interactions.In vitro cellular experiments demonstrated that the two drugs were used as a drug pair to enhance the inhibitory effect on the cell viability of HUVECs.The zebrafish experiments indicated that the toxicity order of herbal couple and single drugs was CR>CR-SR>SR.The results of transgenic zebrafish vascular fluorescence model confirmed that CR-SR and single drugs had anti-angiogenic activity,with the anti-angiogenic activity order of herbal couple and single drugs was CR-SR>SR>CR.The results of qRT-PCR showed that CR-SR drug pairs and single drugs significantly reduced the expression levels of angiogenesis-related genes VEGFA,VEGFR2,EGFR,MMP9,etc.,and had anti-angiogenic effects.Conclusion CR-SR and single drugs had anti-lung cancer effects on multiple identical targets and regulated multiple identical signaling pathways,and their combination had a synergistic effect.The treatment of lung cancer may be through the regulation of angiogenesis-related target VEGFA,VEGFR2,EGFR,etc.,in order to play an anti-angiogenic effect.

Objective To investigate the value of machine learning models based on contrast-enhanced CT radiomics in differentia-ting renal oncocytoma(RO)from chromophobe renal cell carcinoma(chRCC).Methods A total of 65 patients with RO and chRCC confirmed by pathology with complete clinical and imaging data were analyzed retrospectively.The patients were randomly divided into training set(n=45)and test set(n=20)according to a ratio of 7︰3.The tumor boundaries were delineated on the preoperative CT images using 3D Slicer software,and radiomics features were extracted using the Radiomics plugin.Univariate analysis,recursive fea-ture elimination(RFE),least absolute shrinkage and selection operator(LASSO)algorithms were used to select the best radiomics features.Three machine learning models were constructed on the training set and the grid search method was used to select the best combination of hyperparameters.The receiver operating characteristic(ROC)curve,calibration curve and decision curve were used to evaluate the performance of each machine learning model on the training set and test set.Results Random forest model,logistic regres-sion model and support vector machine model can better identify RO and chRCC.In the training set,the area under the curve(AUC)of random forest model and support vector machine model were 0.950[95%confidence interval(CI)0.901-0.998]and 0.955(95%CI 0.908-1.000),respectively,which were higher than the AUC of logistic regression model 0.882(95%CI 0.806-0.956).Statistical differences were found by DeLong test(P<0.05);In the test set,the AUC of random forest model,logistic regression model and support vector machine model were 0.876(95%CI 0.758-0.993),0.883(95%CI 0.768-0.997)and 0.883(95%CI 0.768-0.997),respectively.There was no significant statistical difference in the AUC of each model by DeLong test(P>0.05).The decision curve showed that all three models had significant net clinical benefits.Conclusion The machine learning model based on contrast-enhanced CT radiomics can effectively distinguish RO from chRCC.