Objective:To explore the construction and validation of a nomogram model for predicting poor survival prognosis in patients with clear cell renal cell carcinoma（ccRCC）based on deep learning of pathological images.Methods:This study was an observational cohort study. The original pathological images and clinicopathological data（TCGA cohort）of 378 patients with ccRCC were obtained from the Cancer Genome Atlas Database（TCGA）for model training. A total of 301 patients with ccRCC who underwent surgical treatment at Renji Hospital Affiliated to Shanghai Jiaotong University School of Medicine from January 2010 to December 2020（Renji cohort）and 214 patients with ccRCC who underwent surgical treatment at the First People’s Hospital Affiliated to Shanghai Jiaotong University School of Medicine from January 2012 to December 2018（General cohort）were included for model validation. Their original pathological images and clinical pathological data were collected. A clustering-constrained attention and multi-instance learning method was used to accurately identify sub-regions of the images to classify and extract features of the pathological images. A deep learning-based disease-free survival prognosis prediction model（DL-DFS）was constructed through a weakly supervised learning strategy. The clinical pathological features and DL-DFS were further combined to construct a nomogram model for the clinical prognosis of ccRCC patients. Univariate and multivariate Cox regression analyses were employed to evaluate the independent risk factors for disease-free survival（DFS）. The efficacy of the predictive model were evaluated by the receiver operating characteristic curve（ROC）with area under the curve（AUC），respectively. Survival analysis was conducted using the Kaplan-Meier curve.Results:DL-DFS could accurately predict the DFS status of ccRCC patients in 5 years after surgery. Through ROC analysis in the training cohort，the AUC value reached 0.75（ P < 0.001）. In the Renji cohort and the General cohort，the AUC values were 0.65（ P < 0.001）and 0.81（ P < 0.001），respectively. Through Kaplan-Meier survival analysis，we found that DL-DFS could identify ccRCC patients with high survival risks. The hazard ratio in the training cohort was 3.86（95% CI 2.36-6.30， P < 0.001）. The hazard ratio in the Renji cohort and General cohort were 1.97（95% CI 1.03-3.80， P = 0.009）and 4.66（95% CI 1.80-12.06， P = 0.008），respectively. Univariate and multivariate Cox regression analyses indicated that DL-DFS risk score，tumor grade，and tumor stage could act as prognostic risk factors for patients with ccRCC（ P < 0.05）. Considering that age was a common prognostic risk factor for patients with renal cancer，a nomogram model was constructed by combining the DL-DFS risk score with patient age，tumor grade，and tumor stage. The AUC of this model for predicting the 5-year DFS of ccRCC patients after surgery was 0.87，which was significantly higher than that of DL-DFS（AUC = 0.74），tumor stage（AUC = 0.84），tumor grade（AUC = 0.72），and patient age（AUC = 0.56）in the TCGA cohort（all P<0.05）. In the Renji cohort and the General cohort，the AUC of the nomogram model were 0.78 and 0.86 respectively，which was significantly higher than that of DL-DFS（0.65 and 0.81），tumor stage（0.72 and 0.69），tumor grade（0.64 and 0.77），and patient age（0.56 and 0.63）. Conclusions:In this study a DL-DFS for ccRCC patients was constructed. Then a nomogram model was constructed by combining the DL-DFS risk value with patient age，tumor grade，and tumor stage. This nomogram model demonstrated superior predictive performance compared to DL-DFS alone in evaluating the DFS prognosis of ccRCC patients，which still needs to be further verified in prospective clinical studies.

Objective:To explore the construction and validation of a nomogram model for predicting poor survival prognosis in patients with clear cell renal cell carcinoma（ccRCC）based on deep learning of pathological images.Methods:This study was an observational cohort study. The original pathological images and clinicopathological data（TCGA cohort）of 378 patients with ccRCC were obtained from the Cancer Genome Atlas Database（TCGA）for model training. A total of 301 patients with ccRCC who underwent surgical treatment at Renji Hospital Affiliated to Shanghai Jiaotong University School of Medicine from January 2010 to December 2020（Renji cohort）and 214 patients with ccRCC who underwent surgical treatment at the First People’s Hospital Affiliated to Shanghai Jiaotong University School of Medicine from January 2012 to December 2018（General cohort）were included for model validation. Their original pathological images and clinical pathological data were collected. A clustering-constrained attention and multi-instance learning method was used to accurately identify sub-regions of the images to classify and extract features of the pathological images. A deep learning-based disease-free survival prognosis prediction model（DL-DFS）was constructed through a weakly supervised learning strategy. The clinical pathological features and DL-DFS were further combined to construct a nomogram model for the clinical prognosis of ccRCC patients. Univariate and multivariate Cox regression analyses were employed to evaluate the independent risk factors for disease-free survival（DFS）. The efficacy of the predictive model were evaluated by the receiver operating characteristic curve（ROC）with area under the curve（AUC），respectively. Survival analysis was conducted using the Kaplan-Meier curve.Results:DL-DFS could accurately predict the DFS status of ccRCC patients in 5 years after surgery. Through ROC analysis in the training cohort，the AUC value reached 0.75（ P < 0.001）. In the Renji cohort and the General cohort，the AUC values were 0.65（ P < 0.001）and 0.81（ P < 0.001），respectively. Through Kaplan-Meier survival analysis，we found that DL-DFS could identify ccRCC patients with high survival risks. The hazard ratio in the training cohort was 3.86（95% CI 2.36-6.30， P < 0.001）. The hazard ratio in the Renji cohort and General cohort were 1.97（95% CI 1.03-3.80， P = 0.009）and 4.66（95% CI 1.80-12.06， P = 0.008），respectively. Univariate and multivariate Cox regression analyses indicated that DL-DFS risk score，tumor grade，and tumor stage could act as prognostic risk factors for patients with ccRCC（ P < 0.05）. Considering that age was a common prognostic risk factor for patients with renal cancer，a nomogram model was constructed by combining the DL-DFS risk score with patient age，tumor grade，and tumor stage. The AUC of this model for predicting the 5-year DFS of ccRCC patients after surgery was 0.87，which was significantly higher than that of DL-DFS（AUC = 0.74），tumor stage（AUC = 0.84），tumor grade（AUC = 0.72），and patient age（AUC = 0.56）in the TCGA cohort（all P<0.05）. In the Renji cohort and the General cohort，the AUC of the nomogram model were 0.78 and 0.86 respectively，which was significantly higher than that of DL-DFS（0.65 and 0.81），tumor stage（0.72 and 0.69），tumor grade（0.64 and 0.77），and patient age（0.56 and 0.63）. Conclusions:In this study a DL-DFS for ccRCC patients was constructed. Then a nomogram model was constructed by combining the DL-DFS risk value with patient age，tumor grade，and tumor stage. This nomogram model demonstrated superior predictive performance compared to DL-DFS alone in evaluating the DFS prognosis of ccRCC patients，which still needs to be further verified in prospective clinical studies.

Objective To observe the value of CT radiomics-based nomogram for predicting difference of Lauren types of gastric cancers between endoscopic biopsy and postoperative pathology.Methods Totally 126 patients with gastric cancer diagnosed by surgical pathology were retrospectively analyzed.The patients were divided into concordant group(n=77)and inconsistent group(n=49)according to the concordance between endoscopic biopsy and postoperative pathology results or not,also divided into training set and validation set at the ratio of 2∶1.Clinical predictors were screened,then a clinical prediction model was constructed.Radiomics features were extracted based on venous-phase CT images and screened using L1 regularization.Radiomics models were constructed using 3 machine learning(ML)algorithms,i.e.decision trees,random forests and logistic regression.The nomogram based on clinical and the best ML radiomics model was constructed,and the efficacy and clinical utility of the above models and nomogram for predicting inconsistency of Lauren types of gastric cancers between endoscopic biopsy and postoperative pathology were evaluated.Results Patients'age,platelet count,and arterial-phase CT values of tumors were all independent predictors of inconsistency between endoscopic biopsy and postoperative pathology of Lauren types of gastric cancer.CT radiomics model using random forests algorithm showed better predictive efficacy among 3 ML models,with the area under the curve(AUC)of 0.835 in training set and 0.724 in validation set,respectively.The AUC of clinical model,radiomics model and the nomogram in training set was 0.764,0.835 and 0.884,while was 0.760,0.724 and 0.841 in validation set,respectively.In both training set and validation set,the nomogram showed a good fit and considerable clinical utility.Conclusion CT radiomics-based nomogram had potential clinical application value for predicting inconsistency of Lauren types of gastric cancers between endoscopic biopsy and postoperative pathology.

OBJECTIVETo compare the characteristics of urinary tract infection (UTI) between kidney transplant recipients and non-recipient patients.

METHODSForty-nine kidney transplant recipients with UTI (69 episodes) and 401 non-recipient patients with UTI (443 episodes) admitted in Nanfang Hospital from January 2003 to August 2014 were enrolled in the study. The characteristics of UTI were compared between two groups.

RESULTSIn both groups of UTI, female patients comprised a greater proportion (63.3% and 58.6%) and Escherichia coli was the most common pathogen isolated (37.7% and 34.1%). However, the infection rate of Klebsiella pneumonia in recipients was higher than that in non-recipients (11.6% vs 3.2%, P= 0.001), while the infection rate of Candida albicans was lower (1.5% vs 11.3%, P=0.008) than that in non-recipients. Recipients were likely to develop antibiotic resistance and with a higher recurrence rate than non-recipient patients (38.8% vs 16.7%, P<0.001). Compared to non-recipient UTI patients, the symptoms of urinary irritation in recipient UTI patients were more common. There was higher percentage of neutrophil granulocyte (72.65% ± 1.90% vs 68.59% ± 0.73%, P=0.048), lower proportion of lymphocytes (17.73% ± 1.27% vs 21.28% ± 0.61%, P=0.037), and less platelets [(187.64 ± 10.84) × 10(9)/L vs (240.76 ± 5.26) × 10(9)/L, P<0.01] in recipients than in non-recipient UTI patients.

CONCLUSIONThese results indicate that the characteristics of UTI in kidney transplantation recipients and non-recipients patients are different.

Candida albicans ; isolation & purification ; Escherichia coli ; isolation & purification ; Female ; Humans ; Kidney Transplantation ; Klebsiella pneumoniae ; isolation & purification ; Male ; Transplant Recipients ; Urinary Tract Infections ; epidemiology ; pathology

Objective To analyze the injury features of the wounded in the bullet train crash in July 23,2011 in Wenzhou of Zhejiang and explore treatment experience.Methods A total of 177 cases who were injured or died in the 7.23 Wenzhou bullet train crash were involved in the study.The age,gender,injury types,injury regions and injury severity were analyzed by using Trauma Database System V3.0 ( produced by China Trauma Data Center of Third Military Medical University). Results A total of 137 cases were admitted to our hospital within 24 hours after the bullet train crash in Wenzhou.There were 76 males and 61 females,with no significant difference.Of the 137 cases,136 survived and one died after medical treatment.The major injury types included impact injury,crush injury and heavy weight falling injury.Among the 136 survivors,108 cases were with multiple injuries,accounting for 79.4％.The most common injury region of the survivors was the chest,followed by the head and limb.The minimum trauma index (TI) was 5 points and the maximum 27 points.There were 4 cases(3.7％ )with TI ≤9 points,86 (79.6％) with 10-16 points and 18 (16.7％) with ≥ 17 points.Of all,78 cases (72.2％) had ISS score ＜ 16 points and 30 (27.8％) had ≥16 points.The minimum ISS was 3 points and maximum 75 points.There were 40 deaths including the one died after medical treatment,who were all with multiple injuries,with head injury the main cause.ConclusionsThe causes of the bullet train crashes are complicated,with high incidence of multiple injuries.The injury severity is varied,but the overall situation is not serious.Distribution of common injury regions is significantly different between the wounded and the dead.Injury triage,patient transfer and specialist treatment of the wounded should be carried out appropriately during the initial stage of medical rescue based on those injury features after the bullet train crashes.