论文部分内容阅读
目的建立个体血脂异常患病危险度的预测模型,探讨并评价预测个体血脂异常的新方法。方法选择8914例社区居民流行病学调查资料,按3∶1分为训练集(6686例)与检验集(2228例),分别用于筛选变量、建立预测模型及对模型的检测和评价。应用人工神经网络(ANN)和logistic回归分别建立血脂异常预测模型,受试者工作曲线(ROC)评价预测模型的优劣。结果 ANN预测模型的特异度(64.79%)较低,但灵敏度(94.86%)、约登指数(59.65%)、一致率(81.23%)均优于logistic回归预测模型(特异度=77.49%、灵敏度=53.51%、约登指数=31.00%、一致率=81.23%);ANN预测模型ROC曲线下面积(Az=0.824±0.009)明显大于logistic回归预测模型(Az=0.655±0.012)(P<0.05)。结论在预测个体血脂异常方面,ANN模型较logistic回归模型具有更好的预测判别效能。
Objective To establish a prediction model of the prevalence of dyslipidemia in individuals and to explore and evaluate a new method for predicting the abnormality of dyslipidemia in individuals. Methods The epidemiological survey data of 8,914 community residents were selected and divided into training set (6686 cases) and test set (2228 cases) according to 3: 1, respectively, for screening variables, establishing prediction models and testing and evaluating the models. The prediction models of dyslipidemia were established by artificial neural network (ANN) and logistic regression respectively. The receiver operating curve (ROC) was used to evaluate the quality of the prediction model. Results The specificity of ANN predictive model was lower (64.79%), but the sensitivity (94.86%), Youden index (59.65%) and concordance rate (81.23%) were better than the logistic regression model (specificity = 77.49% = 53.51%, Youden index = 31.00%, concordance rate = 81.23%). The area under the ROC curve of ANN predictive model was significantly higher than that of logistic regression model (Az = 0.824 ± 0.009) . Conclusion The ANN model has better predictive discrimination performance than the logistic regression model in predicting individual dyslipidemia.