| Objectives:This study is designed to build the onset risk prediction model for three kinds of common diabetic complications(diabetic nephropathy,diabetic retinopathy and diabetic foot)by the use of optimized Logistic regression model based on a Systemic reviews and Meta-analysis,and to draw a diagram of individual onset risk evaluation,so as to provide assistance for the individualized therapy of diabetic complications and early detection,diagnosis and treatment for the susceptible population,in an effort to achieve the goal of promoting health,reducing medical cost and controlling the morbidity of diabetic complications.Methods:(1)Systemic reviews and Meta-analysis: based on the established retrieval strategy and inclusion and exclusion criteria,the relevant studies had been retrieved to extract the data,the onset risk factor of three diabetic complications had been systemically reviewed,the value-at-risk of each of their onset factors had been integrated;(2)The building and verification of model: based on the onset risk factors screened from the last step,the optimized Logistic regression model had been substituted to build the onset risk prediction model of three kinds of diabetic complication,and evaluate the validity of model through ROC curve.(3)Drawing the diagram of individual onset risk evaluation : based on the Logistic regression model,the distribution of ratio of P value of dataset to population mean exposure rate was calculated and simulated through digital simulation,thereby the population had been stratified into three risk levels,namely,low risk,moderate risk and high risk,according to level of onset risk of disease,and marked with different colors in order to draw the diagram,which was finally verified through actual data.Results:(1)The Result of Meta analyze:Diabetic nephropathy:A total of 25 studies were included in this study.There were 7 factors into the model,including LDL-C,SBP,FPG,HbAc1,hypertension,diabetes duration and smoking,and their pooled OR were 1.56,1.06,1.40,1.44,1.70,1.20 and 1.93.Diabetic retinopathy:A total of 29 studies were included in this study.There were 7 factors into the model,including HbAc1,TC,mAlb,SBP,,hypertension,diabetes duration and sex,and their pooled OR were 1.67,1.39,,2.48,1.13,1.50,1.53 and 1.73.Diabetic foot:A total of 23 studies were included in this study.There were 6 factors into the model,including eGFR,WBC,HbAc1,age,peripheral neuropathy and peripheral vascular disease,and their pooled OR were 1.84,1.77,1.39,1.07,5.14 and 3.97.(2)The Result of the health risk assessment model:Diabetic nephropathy:0.44 0.06 0.34 0.36 0.53 0.18 0.66Diabetic retinopathy0.51 0.33 0.91 0.12 0.41 0.43 0.55Diabetic foot:0.61 0.57 0.33 0.07 1.64 1.38(3)The Result of Validation:The result showed that,the area under the curve(AUC)of ROC in diabetic nephropathy was 0.662(0.591-0.733);the AUC in diabetic retinopathy was 0.702(0.631-0.772);and the AUC in diabetic foot was 0.841(0.773-0.909).(4)The Result of verification of risk evaluation diagram: 362 cases of type-2 diabetic patient from the endocrinology department of Chongqing Southwest Hospital had been divided according to the low,moderate and high risk threshold in the risk evaluation diagram.The result showed that,the area under the curve(AUC)of ROC in diabetic nephropathy was 0.678(0.609-0.748);the AUC in diabetic retinopathy was 0.649(0.575-0.722);and the AUC in diabetic foot was 0.859(0.802-0.916).Conclusion:This study has adopted the optimized Logistic regression model to build a onset risk prediction model for three kinds of diabetic complication based on a Systemic reviews and Meta-analysis,and for the first time has drawn its onset risk evaluation diagram through digital simulation on the basis of the established model,producing a good predictive effect.The findings of this study is of a certain referential significance to the prediction of the individual onset risk of three diabetic complications,providing a new thought for the establishment of model for other chronic diseases. |
PDF Full Text Request