2016-2020年乌鲁木齐市新市区全民健康体检人群血脂异常检出情况及时间序列预测研究

Study on detection and time series prediction of dyslipidemia in health examination in Xinshi District of Urumqi from 2016 to 2020

目的分析2016-2020年乌鲁木齐市新市区全民健康体检人群血脂异常检出情况及影响因素,运用ARIMA时间序列模型对血脂异常情况进行拟合和预测。方法获取2016-2020年新市区年龄≥18岁全民健康体检人群血脂异常的相关数据及报表,分析5年间健康体检人群血脂异常的检出情况,以及血脂异常的影响因素,利用时间序列ARIMA模型对2016年1月-2019年12月的血脂异常检出率进行建模,选出最优模型对2020年1-12月血脂异常的检出率进行拟合及预测。结果年龄、性别(除2016年)、吸烟情况(除2019年)、饮酒频次(除2019年)、文化程度和体质指数是乌鲁木齐市新市区健康体检人群血脂异常的影响因素(P均<0.05)。在18~69岁年龄区间,血脂异常检出率随着年龄的增大而升高;除2016年外,男性的血脂异常检出率高于女性;有吸烟史者的血脂异常检出率较不吸烟者高;每天饮酒者的血脂异常检出率较高;文化程度大专及以上者的血脂异常检出率较低;超重和肥胖者的血脂异常检出率升高。通过赤池信息量(Akaike information criterion,AIC)和贝叶斯信息量(Bayesian information criterion,BIC)最小原则得出,ARIMA(1,1,1)(0,1,1)12为最优模型。该模型对2020年1-12月的血脂异常预测的平均绝对百分比误差为15.84%,除外疫情影响的月份,实际检出率均在95%的置信区间。结论2016-2020年新市区全民健康体检人群中男性、40岁以上、有吸烟史、每天饮酒、文化程度较低、超重和肥胖者血脂异常的检出率较高。ARIMA(1,1,1)(0,1,1)12模型能较好地对血脂异常的检出率进行拟合和短期预测,对新市区血脂异常防治有一定的指导意义。

Objective To analyze the condition and influencing factors of dyslipidemia in health examination of all people in Xinshi District of Urumqi from 2016 to 2020,and to use autoregressive intergrated moving average(ARIMA)time series model to fit and predict the condition of dyslipidemia.Methods To obtain the relevant data and reports of dyslipidemia among all healthy people aged≥18 years old in the Xinshi District from 2016 to 2020,and to analyze the detection of dyslipidemia among the healthy people in the past 5 years,as well as the influencing factors of dyslipidemia.The time series ARIMA model was used to model the detection rate of dyslipidemia from January 2016 to December 2019,and the best model was selected to fit and predicted the detection rate of dyslipidemia in the 12 months of 2020.Results Age,sex(except 2016),smoking status(except 2019),drinking frequency(except 2019),education level and body mass index were the influencing factors of dyslipidemia in healthy people in Xinshi District of Urumqi(all P<0.05).In the age range of 18-69 years,the detection rate of dyslipidemia was increased with the increasing of age.Except in 2016,the detection rate of dyslipidemia in males were higher than that in females.The detection rate of dyslipidemia in smokers was higher than that in non-smokers.The detection rate of dyslipidemia was higher in people who drank alcohol every day.The detection rate of dyslipidemia was lower in those with college education or above.The detection rate of dyslipidemia was increased in overweight and obese patients.According to the minimum principle of Akaike information criterion(AIC)and Bayesian information criterion(BIC),ARIMA(1,1,1)(0,1,1)12 was the optimal model.The average relative error percentage of the model for the prediction of dyslipidemia from January to December 2020 was 15.84%,except for the months affected by the epidemic,and the actual detection rate was in the 95%confidence interval.Conclusion The detection rate of dyslipidemia was higher in male,over 40 years old,smoking history,drinking alcohol every day,low education level,overweight and obese people in the national health examination population in Xinshi District from 2016 to 2020.ARIMA(1,1,1)(0,1,1)12 model can better fit and predict the detection rate of dyslipidemia in the short term,which has certain guiding significance for the prevention and treatment of dyslipidemia in Xinshi District.