The relationship between impaired fasting blood glucose and blood indicators based on random forest: a follow-up cohort risk prediction study
10.3760/cma.j.cn211501-20201229-04975
- VernacularTitle:基于随机森林的空腹血糖受损与血液指标的关系：一项随访队列风险预测研究
- Author:
Hongyu ZHANG
1
;
Wenqi HU
;
Cuiqing XU
Author Information
1. 山东大学附属千佛山医院健康管理中心，济南 250014
- Keywords:
Fasting blood glucose;
Blood test;
Risk prediction;
Random forest
- From:
Chinese Journal of Practical Nursing
2021;37(23):1767-1774
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the relationship between blood test indicators and their changes and impaired fasting blood glucose, and to conduct risk prediction research.Methods:This study selected people who underwent a physical examination at a health management center in Jinan from 2018 to 2019, with an interval of 6 months or more between the two physical examinations, and had no history of diabetes and normal fasting blood sugar during the first year of the physical examination. We took the fasting blood glucose level in the second year as the dependent variable, and took the general examination and blood test indicators in the first year as well as the changes of these indicators as independent variables to construct the model. First, the random forest model was used to screen the independent variables, and then the logistic regression model is fitted to calculate the effect values of the independent variables.Results:A total of 4 416 subjects were enrolled in this study. After a 1-year follow-up, 3.45% (133/3 851) subjects developed prediabetic blood glucose, and 0.42% (16/3 851) developed diabetic blood glucose. The sensitivity of the model using blood test indicators to predict the risk of impaired fasting blood glucose after 1 year was 67.1%, and the specificity was 100.0%. The risk of impaired fasting blood glucose after 1 year was high in the population with older age ( OR value was 1.024, 95% CI 1.014-1.035), higher systolic blood pressure ( OR value was 1.018, 95% CI 1.011-1.025), higher body mass index ( OR value was 1.079, 95% CI 1.036-1.125), higher uric acid ( OR value was 1.003, 95% CI 1.002-1.005), higher glutamyl transpeptidase ( OR value was 1.006, 95% CI 1.003-1.010), lower creatinine ( OR value was 0.971, 95% CI 0.957-0.985), lower AST/ALT ( OR value was 0.614, 95% CI 0.411-0.917)) and lower estimated glomerular filtration rate ( OR value was 0.972, 95% CI 0.958-0.988). The sensitivity of the model using the changes of blood test indicators to predict the risk of impaired fasting blood glucose after 1 year was 65.2%, and the specificity was 99.7%. The risk of impaired fasting blood glucose is high in the population with increased triglycerides ( OR value was 1.152, 95% CI 1.057-1.257), increased alkaline phosphatase ( OR value was 1.021, 95% CI 1.012-1.030), increased albumin ( OR value was 1.087, 95% CI 1.029-1.148), increased aspartate aminotransferase ( OR value was 1.006, 95% CI 1.003-1.009), increased weight ( OR value was 1.049, 95% CI 1.019-1.081) within 1 year. Conclusions:The possibility of impaired fasting blood glucose after 1 year is high in the population with older age, higher systolic blood pressure, higher body mass index, higher uric acid, higher glutamyl transpeptidase, lower creatinine, lower AST/ALT and lower estimated glomerular filtration rate. The possibility of impaired fasting blood glucose is high in the population with increased triglycerides, increased alkaline phosphatase, increased albumin, increased aspartate aminotransferase and increased weight with in 1 year. When the above situation occurs in the physical examination, the blood glucose should be monitored and timely intervention should be taken to prevent the development of diabetes.
