人工神经网络模型在2型糖尿病患病风险预测中地应用.docVIP

人工神经网络模型在2型糖尿病患病风险预测中的应用 * 王重建, 李星, 王玲, 郭奕瑞，王高帅, 李文杰 （郑州大学公共卫生学院营养与食品卫生学系, 河南 郑州 450001） [摘要　目的　。方法 8640名居民流行病学调查，按3:1分为训练集（6例）与检验集（2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回归分别建立预测模型，用受试者工作曲线(ROC)评价预测模型的。结果　ANN预测模型的灵敏度（%）、特异度（%）、（%）、（%）优于logistic回归预测模型（灵敏度=%、特异度=%、=19.94%、=94.50%）ANN预测模型曲线下面积（Az=0.±0.015）明显大于logistic回归预测模型（Az=0.±0.021）。结论　，ANN模型较logistic回归模型。【关键词】　；人工神经网络；logistic回归；危险因素；模型Application of artificial neural networks to predict individual health risk of type 2 diabetes mellitus WANG Chong-jian, LI Xing, WANG Ling, GUO Yi-rui, WANG Gao-shuai, LI Wen-jie (Department of nutrition and food hygiene, School of Public Health, Zhengzhou University, Zhengzhou, 450001) 【Abstract】Objective To explore the potential application of artificial neural network (ANN) on , and then to develop and evaluate an effective and inexpensive prediction approach. Methods A cross-sectional survey was conducted. Of 8640 subjects who met inclusion criteria, 75% (N1=6480) were randomly selected to provide training set for constructing artificial neural network (ANN) and multivariate logistic regression (MLR) models. The remaining 25% (N2=2160) were assigned to validation set for performance comparisons of the ANN and MLR models. Predictive performance of different models was analyzed by the receiver operating characteristic (ROC) curve using the validation set. Results For ANN model, the sensitivity, specificity, positive and negative predictive value for identifying T2DM were 86.93%, 79.14%, 31.86%, and 98.18%, respectively, while MLR model were only 62.81%, 71.70%, 19.94%, and 94.50%, respectively. Area under the ROC curve (AUC) value for identifying T2DM when using the ANN model was 0.891±0.015, showing more accurate predictive performance than the MLR model (AUC =0.742±0.021) (P=0.0001). Conclusion The ANN model has a better discriminated performance than logistic regression in the prediction of individual health risk of T2DM. 【Key words】 Artificial neural network; Lo