Objective To explore the feasibility of a deep learning model based on early fundus lesions without manual segmentation in pan-retinal images for predicting diabetic kidney disease(DKD)and evaluating the enhancing effects of different binocular fusion strategies.Methods A retrospective cohort of 353 patients with type 2 diabetes mellitus(T2DM)admitted to the Endocrinology Department of Jiangdu People's Hospital Affiliated to Yangzhou University between December 2022 and March 2024 was analyzed.Patients were divided into DKD(n=114)and non-diabetic kidney disease(NDKD)(n=239)group based on the presence of DKD.First,a U-Net-based pre-trained automatic segmentation model was developed to process panoramic fundus retinal images.Subsequently,left and right eye deep learning models were constructed using ResNet152 under a five-fold cross-validation framework(70%training,30%validation).Three binocular fusion strategies were implemented:result fusion,feature fusion,and image fusion models.Model performance was evaluated using the area under the receiver operating characteristic(ROC)curve(AUC).DeLong test was used to compare AUC differences among models,while net reclassification index(NRI)and decision curve analysis(DCA)were used to assess clinical utility.Results Six prediction models were developed:clinical parameter model,left fundus model,right fundus model,binocular image fusion model,binocular result fusion model,and binocular feature fusion model.The Transformer-based binocular feature fusion model achieved the highest AUC in both training and validation sets(0.864 and 0.658,respectively).DeLong tests revealed significant AUC superiority of the Transformer model over the other five models in the training set(all P<0.001),though no significant differences were observed in the validation set(all P>0.05).NRI analysis showed negative values for all comparisons with the Transformer model(training set:-0.255,-0.244,-0.289,-0.426,-0.163;validation set:-0.060,-0.016,-0.028,-0.105,-0.033,respectively),indicating its optimal predictive performance.DCA further demonstrated greater net benefit for the Transformer-based fusion model.Conclusions The deep learning model constructed using early fundus lesions without manual segmentation in pan-retinal images can predict DKD.The Transformer-based fusion strategy present the best performance,providing a novel approach for further optimization and development of tools to predict DKD in the future.

Objective To explore the feasibility of a deep learning model based on early fundus lesions without manual segmentation in pan-retinal images for predicting diabetic kidney disease(DKD)and evaluating the enhancing effects of different binocular fusion strategies.Methods A retrospective cohort of 353 patients with type 2 diabetes mellitus(T2DM)admitted to the Endocrinology Department of Jiangdu People's Hospital Affiliated to Yangzhou University between December 2022 and March 2024 was analyzed.Patients were divided into DKD(n=114)and non-diabetic kidney disease(NDKD)(n=239)group based on the presence of DKD.First,a U-Net-based pre-trained automatic segmentation model was developed to process panoramic fundus retinal images.Subsequently,left and right eye deep learning models were constructed using ResNet152 under a five-fold cross-validation framework(70%training,30%validation).Three binocular fusion strategies were implemented:result fusion,feature fusion,and image fusion models.Model performance was evaluated using the area under the receiver operating characteristic(ROC)curve(AUC).DeLong test was used to compare AUC differences among models,while net reclassification index(NRI)and decision curve analysis(DCA)were used to assess clinical utility.Results Six prediction models were developed:clinical parameter model,left fundus model,right fundus model,binocular image fusion model,binocular result fusion model,and binocular feature fusion model.The Transformer-based binocular feature fusion model achieved the highest AUC in both training and validation sets(0.864 and 0.658,respectively).DeLong tests revealed significant AUC superiority of the Transformer model over the other five models in the training set(all P<0.001),though no significant differences were observed in the validation set(all P>0.05).NRI analysis showed negative values for all comparisons with the Transformer model(training set:-0.255,-0.244,-0.289,-0.426,-0.163;validation set:-0.060,-0.016,-0.028,-0.105,-0.033,respectively),indicating its optimal predictive performance.DCA further demonstrated greater net benefit for the Transformer-based fusion model.Conclusions The deep learning model constructed using early fundus lesions without manual segmentation in pan-retinal images can predict DKD.The Transformer-based fusion strategy present the best performance,providing a novel approach for further optimization and development of tools to predict DKD in the future.

Under the organization of Renji Hospital, Shanghai Jiao Tong University School of Medicine, a specialized disease database of inflammatory bowel disease (IBD) cohort was deployed, and a brief introduction of the database was made in this article. The IBD data set was established by referring to domestic and foreign standards. Through data extraction, cleaning, normalization and other information processing technologies, data from multi‑source heterogeneous platform were arranged to form a specialized major disease database of IBD cohort and the efficiency and quality of data collection in clinical practice, teaching and scientific research were guaranteed. The display and personalized export capacities of the database can promote the researches on IBD and assist the clinical decision‑making. It provides not only efficient, comprehensive and reliable research‑level data support for scientific research, but also a precise guidance for diagnosis and treatment of the disease. Furthermore, it can excavate the potential clinical principles based on medical big data.

Objective To study formation of ammonion-magnesium phosphate crystals in urine with bacteria growing and provide guidance for cilinical prevention of urinary calculi.Methods Bacterial culturefluid of Escherichia coli,Proteus mirabilis,Pseudomonas aeruginosa,Klevsiella pneumoniae,Enterococcus in urine was examined directly under the ultrahigh sensitive microscpcope system for ammonion-magnesium phosphate crystasl.The number of ammonion-magnesium phosphate crystasl was measured when the 24th and the 48th hour.Results Ammonion-magnesium phosphate crystasl were observed from the culture fluid without ammonion magnesium phosphate crystasl.The rate of male formation was higher than that of female.Ammonion-magnesium phosphate crystals in culture fluid of Proteus mirabilis was the highest,Pseudomonasaeruginosa was the second,the third was Klebsiella pneumoniae,and there was formed 1 case in 2 ml culturefluid of enterococcus,and 2 cases of formation in 5 ml culturefluid of Escherichia coli.The crystals formed were the most unformed feather crystals,followed by cubic and square cylinders,an d the envelope like crystals were the least.Conclusion Bacteria with urease play a significant role in ammonion-magnesium phosphate crystasl formed,Proteus mirabilis is the main pathlogen.

Inflammatory bowel disease (IBD)includes Crohn’s disease (CD)and ulcerative colitis (UC).The differential diagnosis between CD and UC mainly depends on clinical symptoms,endoscopy,pathological biopsy,laboratory and imaging examinations.In recent years,studies with a variety of IBD-related biomarkers develop rapidly because of its non-invasiveness,simple and easily acceptable.With the development of genome-wide association study (GWAS),great progress has been achieved in studies of gene mutations and susceptibility genes related with CD and UC,which provides new approach for diagnosis of the disease.This article reviewed the advances in studies on serum biomarkers and susceptibility genes in differential diagnosis of IBD.

Objective To investigate antimicrobial resistance and pathogen in hebei antibacterial resistance investigation net in 2012.Methods Antimicrobial susceptibility test was detected by Kirby-Bauer method or broth dilution test.Results were analyzed according to CLSI 2010 breakpoints.WHONET 5.5 software was used to analyze the data.Results A total of 10 504 clinical isolates were collected in 2012,of which gram negative bacilli and gram positive cocci accounted for 76.2%, 23.8%,respectively.The most common pathogen in gram-negative rod was E.coli,K.pneumoniae,P.aeruginosa, A.baumanii and E.cloacae respectively.The most common pathogen in gram-positive cocci was S.aureus,E.facium,E-.faecalis,S.pneumoniae and S.epidermidis.ESBL rate of E.coli and K.pneumoniae was 66.5 and 46.7%.The resistant rate of E.coli,K.pneumoniae,E.cloacae to imipenem was 0.1%,0.5%,8.9% and to meropenem was 0.1%,0.6%,4.2%, respectively.P.aeruginosa was resistant to imipenem and meropenem were 38.9% and 32.3%.A.baumanii was resistant to imipenem and meropenem were 5 6.5% and 5 9.7%.Methicillin-resistant strains accounted for an average of 5 7.5% in S.aureus and 87.3% in coagulase negative staphylococcus.Staphylococcus was still susceptible to minocycline and chloram-phenicol.No staphylococcal strains were found resistant to vancomycin,linezolid.But a few coagulase negative staphylococcal strains were resistant to teicoplanin.Conclusion Surveillance of antimicrobial agents played an important role in controlling hospital infection.