论文部分内容阅读
目的建立基于BP神经网络的老年男性保健人群缺血性心脑血管病(ischemic cardiovascular disease,ICVD)的发病预测模型。方法以某保健医院所有老年男性保健对象2003年5月的体检资料为基线数据,剔除基线前已患有ICVD者。收集历年住院资料、问卷调查资料、电话回访资料,观察截止到2010年10月的ICVD发病及死亡情况。按照4∶1的比例将基线人群划分为训练组和测试组,利用MATLAB 7.0软件的神经网络工具箱构建训练组的BP神经网络模型,用测试组加以验证,并与传统的COX比例风险回归模型进行比较。结果 BP人工神经网络模型的判别能力、群体水平的预测能力方面优于COX比例风险回归模型。结论利用BP神经网络进行疾病预测,能获得更好的预测效果,并且对资料的类型、分布不作任何要求,是一种较好的疾病预测方法。
Objective To establish a predictive model of ischemic cardiovascular disease (ICVD) in the elderly male health care based on BP neural network. Methods According to the physical examination data of all elderly male health care subjects in a health care hospital in May 2003, the baseline data were obtained before the baseline had ICVD. Collect historical data of hospitalization, questionnaire survey and telephone return visit to observe the incidence and death of ICVD as of October 2010. According to the 4: 1 ratio, the baseline population was divided into training group and testing group. The BP neural network model of training group was constructed by using the neural network toolbox of MATLAB 7.0 software, verified by the test group, and compared with the traditional COX proportional hazards regression model Compare. Results BP neural network model discriminant ability, population level prediction ability is superior to the COX proportional hazards regression model. Conclusions It is a good method to forecast the disease by using BP neural network to get a better prediction result without any requirement on the type and distribution of the data.