ABSTRACT
Objective To establish simple screening models for nonalcoholic fatty liver disease (NAFLD) in the adult Blang population. Methods Based on the survey data of metabolic diseases in the Blang people aged 18 years or above in 2017, 2993 respondents were stratified by sex and age (at an interval of 5 years) and then randomly divided into modeling group with 1497 respondents and validation group with 1496 respondents. Related information was collected, including demographic data, smoking, drinking, family history of diseases and personal medical history, body height, body weight, waist circumference, and blood pressure, and related markers were measured, including fasting plasma glucose, 2-hour postprandial plasma glucose or blood glucose at 2 hours after glucose loading, triglyceride, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma-glutamyl transpeptidase. The chi-square test was used for comparison of categorical data between two groups. Logistic regression analysis was used to establish the screening model. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, positive predictive value, and negative predictive value were used to evaluate the screening performance of established models versus existing models in the study population, and the DeLong method was used for comparison of AUC. Results Three screening models for NAFLD were established based on physical and biochemical measurements, i.e., simple noninvasive model 1 (age, body mass index, and waist circumference), noninvasive model 2 with the addition of blood pressure, and model 3 with the combination of hematological parameters (diabetes and ALT/AST). In the modeling group, the three models had an AUC of 0.881 (95% confidence interval [ CI ]: 0.864-0.897), 0.892 (95% CI : 0.875-0.907), and 0.894 (95% CI : 0.877-0.909), respectively, and there was a significant difference between model 1 and models 2/3 ( P =0.004 0 and P 0.05). Based on the overall consideration of screening performance, invasiveness, and cost, the simple noninvasive model 1 was considered the optimal screening model for NAFLD in this population. Model 1 had the highest Youden index at the cut-off value of 5 points, and when the score of ≥5 points was selected as the criteria for NAFLD, the model had a sensitivity of 86.5%, a specificity of 79.7%, a positive predictive value of 50.3%, and a negative predictive value of 96.1% in the modeling group and a sensitivity of 85.6%, a specificity of 80.6%, a positive predictive value of 51.7%, and a negative predictive value of 95.8% in the validation group. Conclusion The NAFLD screening models established for the adult Blang population based on age and obesity indicators have relatively higher sensitivity, specificity, and negative predictive value, and this tool is of important practical significance for the intervention of NAFLD and its closely related metabolic diseases in this population.
ABSTRACT
Objective To study the efficiency and accuracy of artificial intelligence (AI) system based on fundus photograph in diabetic retinopathy(DR)screening,and evaluate the clinical application value of AI system. Methods A diagnostic trial was adopted. Total of 13683 color fundus photos were collected in Zhaoqing Gaoyao People's Hospital from March,2017 to November,2018. The AI system for DR (ZOC-DR-V1) was established,based on transfer learning + NASNet algorithm,by training 4465 precisely labeled fundus images (2510 normal,and 1955 with any stage of DR). One thousand confirmed fundus images (300 normal and 700 with any stage of DR),diagnosed by AI ( AI group ) and doctors ( 3 ophthalmologist doctors and 3 endocrinologist doctors ) ( doctor group ) , respectively. Ophthalmologist group and endocrinologist group were both composed of primary,intermediate and senior physicians. The mean reading time of each image and the total time of 1000 images were recorded. The accuracy and efficiency of AI system and doctor groups were compared. The reading process was divided into two stages. The diagnostic coincidence rate and the average reading time of each group between the two parts were calculated and compared. This study protocol was approved by Ethic Committee of Zhongshan Ophthalmic Center, Sun Yat-sen University (No. 2017KYPJ104). Results After training,the diagnostic coincidence rate of AI system (ZOC-DR-V1) in test set was 94. 7％,AUC was 0. 994. In this "man-machine to war",the diagnostic coincidence rate of primary,intermediate and senior endocrinologist was 94. 0％,91. 4％ and 93. 4％;the diagnostic coincidence rate of primary,intermediate and senior ophthalmologist was 92. 7％,94. 4％ and 95. 6％;the diagnostic coincidence rate of AI system was 95. 2％. There was no difference in the diagnostic coincidence rate between AI system and senior ophthalmologist ( P = 0. 749 ) . The mean reading time of each image of primary, intermediate and senior endocrinologists was (4. 63±1. 87),(3. 74±3. 47) and (5. 71±3. 47) seconds,and the total time of 1000 images of primary,intermediate and senior endocrinologists was 1. 29,1. 04 and 1. 58 hours;the mean reading time of each image of primary,intermediate and senior ophthalmologists was ( 7. 25 ± 6. 58 ) , ( 5. 18 ± 5. 01 ) and ( 5. 18 ± 3. 47 ) seconds,and the total time of 1000 images of primary,intermediate and senior endocrinologists was 2. 02,1. 44 and 1. 44 hours;the mean and total time of AI system was (1. 62±0. 67) seconds and 0. 45 hours,and the reading time of AI system was significantly shorter than that of the doctor groups (all at P=0. 000). The diagnostic coincidence rates between previous and posterior part of primary endocrinologist, primary and intermediate ophthalmologist were significantly different (χ2=11. 986,6. 517,10. 896;all at P<0. 05),and the mean reading time in the posterior part was significantly shorter than that in the previous part of intermediate and senior endocrinologist and primary ophthalmologist (t=4. 175,8. 189,5. 160;all at P<0. 01). While the reading time of AI system remained stable throughout the process(χ2=3. 151,P=0. 103;t=0. 038,P=0. 970). Conclusions The ophthalmic AI system based on fundus images has a good diagnostic efficiency,and its diagnostic coincidence rate can compare with senior ophthalmologist,providing a new method and platform for large-scale DR screening.
ABSTRACT
Objective To investigate a diabetic retinopathy ( DR ) detection algorithm based on transfer learning in small sample dataset. Methods Total of 4465 fundus color photographs taken by Gaoyao People ' s Hospital was used as the full dataset. The model training strategies using fixed pre-trained parameters and fine-tuning pre-trained parameters were used as the transfer learning group to compare with the non-transfer learning strategy that randomly initializes parameters. These three training strategies were applied to the training of three deep learning networks:ResNet50,Inception V3 and NASNet. In addition,a small dataset randomly extracted from the full dataset was used to study the impact of the reduction of training data on different strategies. The accuracy and training time of the diagnostic model were used to analyze the performance of different training strategies. Results The best results in different network architectures were chosen. The accuracy of the model obtained by fine-tuning pre-training parameters strategy was 90. 9％,which was higher than the strategy of fixed pre-training parameters (88. 1％) and the strategy of randomly initializing parameters ( 88. 4％) . The training time for fixed pre-training parameters was 10 minutes,less than the strategy of fine-tuning pre-training parameters ( 16 hours ) and the strategy of randomly initializing parameters (24 hours). After the training data was reduced,the accuracy of the model obtained by the strategy of randomly initializing parameters decreased by 8. 6％ on average,while the accuracy of the transfer learning group decreased by 2. 5％ on average. Conclusions The proposed automated and novel DR detection algorithm based on fine-tune and NASNet structure maintains high accuracy in small sample dataset,is found to be robust,and effective for the preliminary diagnosis of DR.
ABSTRACT
Objective To investigate the prevalence and associated risk factors of diabetes and prediabetes in Blang ethnic adults in Menghai county. Methods A cross-sectional survey including 3 365 Blang ethnic adults (aged 18 and above from 5 administrative villages) was conducted from February 2017 to March 2017 in Menghai county. A questionnaire, physical examination, and blood assays were included in the survey. Finally,a total of 3 237 adults with complete data were selected into this analysis. Results The standardized prevalence of diabetes and prediabetes in Blang ethnic adults were estimated based on the sixth national census in 2010. According to the 1999 WHO criteria, the overall standardized prevalence of diabetes and prediabetes were 8.5% (men: 10.2%, women: 6.8%) and 16.1% (men: 18.0%, women: 14.1%), in which the standardized prevalence of newly diagnosed diabetes among the total population was 7.3% (men: 8.7%, women: 5.8%). Multivariable multinominal logistic regression analyses showed that age, hypertension, hypertriglyceridemia, and central obesity were significantly positively associated with both diabetes and prediabetes, with the corresponding odds ratios of 1.74 and 1.37, 2.39 and 2.02, 2.30 and 1.34, 2.55 and 1.73, respectively. Conclusion The prevalence of diabetes is relatively high in Blang ethnic adults in Menghai county. Improving knowledge of diabetes among the local population is one of key steps in the prevention of diabetes.
ABSTRACT
Objective To compare the efficacy of early and delayed removal of debridement implant for infection after internal fixation of tibial fracture.Methods A retrospective case control study was conducted on the clinical data of 27 patients with tibial fractures who received plate or screw internal fixation admitted to the East Hospital Affiliated to Tongji University from March 2005 to September 2016.There were 21 males and six females,aged 18-81 years [(41.6 ± 14.3)years].According to the treatment methods,the patients were divided into the delayed implant removal group (Group A,10 patients) and the early implant removal group (Group B,17 patients).Group A was given debridement and anti infection treatment followed by continuous dressing change,and the implant was removed after the fractures were healed.Group B was given debridement and implant removal after one month of anti infection treatment and continuous dressing change when the infection was not clearly controlled.Patients with stable fracture ends were given only negative pressure closed drainage (VSD),and those with instable fracture ends were given external fixation and VSD.The time from infection to implant removal,the time of infection control,the fracture nonunion rate,the chronic bone infection rate,the knee joint function score of the American Hospital for Special Surgery (HSS),and the American Orthopedic Foot and Ankle Society (AOFAS) ankle hindfoot scale were compared between the two groups.Results All patients were followed up for 13-47 months,with the average of 28.4 months.There were significant differences between Group A and Group B in terms of the time from infection to implant removal [(49.9 17.1) weeks ∶ (19.3 ± 9.2) weeks],the time of infection control [(85.3 ±78.3)days∶ (6.3 ±2.8)days],fracture nonunion rate (30％ ∶ 0),and the chronic osteomyelitis incidence (30％ ∶ 0) (all P ＜0.05).No significant differences were found in HSS knee joint function score and AOFAS ankle hindfoot scale between the two groups (both P ＞ 0.05).Conclusion For patients with postoperative infection after internal fixation for tibial fracture,early thorough debridement and implant removal can quickly control the infection and reduce the incidence of nonunion and osteomyelitis.
ABSTRACT
Objective To investigate the correlation between acute gastrointestinal injury severity and intestinal microenvironment in rabbits with severe multiple trauma.Methods A total of 60 New Zealand white rabbits were enrolled and randomly assigned into the experimental group (48 rabbits) and control group (12 rabbits).In experimental group,the models of traffic-induced injuries were successful made by using self-made small gravity traction colliders.There were nine rabbits with craniocerebral injury combined with damage of liver and spleen,three with four extremity fractures combined with damage of liver and spleen,12 with rib fractures combined with damage of lungs and pleural effusion,11 with epidural hematoma,contusion and laceration of brain as well as fractures of four extremities and pelvis,and nine with multiple fractures.The injuries had met the criteria of multiple severe traumas according to the injury severity score (ISS).The control group had similar condition with experimental group except for participation in injury model.The parameters at time points of 6,12,24 and 48 h after injury were observed.The parameters were:(1)Enzyme linked immunosorbent assay (ELISA) and high pressure liquid chromatography (HPLC) were used to detect the levels of diamine oxidase (DAO) and the ratio of lactulose to mannitol in urine in order to evaluate the permeability of intestinal mucosal barrier.(2) The small intestinal propulsive rate was detected by carbon pushing mcthod.The interstitial cells of Cajal in the snall intestine wall were observed by transmission electron microscopy.The expression of C-kit was detected by Western blot.By these means,the dynamic function of intestinal mucosal barrier was evaluated.(3) The damage degree of intestinal nucosal barrier was evaluated by pathological observation and Chiu score in the end of small intestine and colon.(4)The intestinal bacterial translocation was evaluated by intestinal microflora culture,mesenteric lymph nodes checking as well as translocation examination of liver and spleen.Results (1) The level of DAO plasma and urine lactulose/mannitol ratio at 6 h after injury increased to varied degrees,and reached the peak during 12-24 h,the correlation analysis of which showed that plasma DAO levels were positively correlated with AGI grade at 6,12 and 24 h (r =0.486 3,0.493 3,0.477 6,P ＜ 0.05).The ratio of urinary lactulose/ mannitol excretion was only positively correlated with AGI grade at 6 and 12 h (r =0.478 5,0.497 2,P ＜ 0.05).(2) The gastrointestinal transmission rate in the experimental group was (48.2 ± 5.2) ％,and that of the control group was (60.3 ± 3.0) ％.The gastrointestinal transmission rate of model group rabbits was significantly longer than that of control group (P ＜ 0.05).(3) The interstitial cells of Cajal decreased dramatically during 12-24 h after the injury,and the expression level of C-kit protein decreased consistently with the increase of AGI grade.(4) The intestinal and colonic mucosa tissues of the experimental group showed obvious necrosis at 12-24 h after injury,and the Chiu score increased gradually with the increase of AGI grade under light microscope.(5) The Enterobacter and Enterococcus in the experimental group increased significantly compared with that in the control group.But the numbers of Bifidobacterium,Lactobacillus,Lactobacillus,Bacteroides as well as the ratio of Bifidobacterium to Enterobacteriaceae significantly decreased(P ＜0.05).(6)The organ bacterial translocation rates were 24％,42％ and 62％ after injury in experimental group (P ＜0.05).Conclusions Acute gastrointestinal injury may occur early after severe multiple trauma,and the injury severity is closely related to the change of intestinal barrier function and bacterial translocation.Early attention and active correction of the change of intestinal microenvironment are of great importance for treatment of multiple trauma.
ABSTRACT
Objective To establish an orthotopic rat model of bladder cancer induced by N-methyl.nitrosourea (MNU)and to evaluate the diagnostic value of CT(computed tomography)scanning in this experimental model.Methods 50 SD rats were randomly divided into the control group(group A)with 15 rats and experimental group(group B)with 35 rats.Each rat of group B was treated with 2 mg MNU per dose every other week,totally 4 doses,by per urethra administration.Meanwhile,each rat of group A was treated with normal saline.Then,at the 14th week,all the rats were evaluated by CT scanning and pathological examination.Results Abnormal changes were detected in each of the 28 rats in group B by CT scanning,and manifested as local mass,thickening bladder wall accompanied with heterogeneous density.Bladder cancer was diagnosed by pathology.However,no bladder tumor was detected by CT scanning and pathologicalexamination in group A.Conclusion A rat model of orthotopic bladder cancer can be established by per urethra administration of MNU.CT scanning is a reliable diagnostic technique concerning this model.Furthermore,this technique can render US a more suitable rat model for further experimental studies on chemotherapy and radiotherapy of bladder Cancer.