基于神经网络的智慧社区居民高血压预测研究

Study on Prediction of Hypertension of Residents in Intelligent Community Based on Neural Network

随着智慧社区的实践建设和发展,传统的社区服务已无法满足居民的要求,智慧医疗作为社区服务的重要内容,更必不可少。面对当代社区居民生活习惯不良,身体检查不及时而导致慢性病病例日益增多的情况,建立了针对高血压疾病的BP神经网络(BPNN)预测模型。针对智慧社区数据采集不完备、数据丢失等导致的单值缺失问题,提出了基于BPNN预测插补的算法,以已知健康信息预测插补出缺值信息,再将补全后的数据预测出居民得高血压的风险,即可对含缺值的样本进行高血压预测的方法。实验结果显示BPNN插补法比传统的插补法的准确度更高,误差为5.3%,且插补后数据应用于高血压预测也效果更优,正确率为93%,即该模型可对不完备数据样本进行高血压预测,在居民体检前可提供预测结果,节省医疗资源,为社区居民提供一定的医疗保障服务。

With the practical construction and development of smart communities, traditional community services have been unable to meet the requirements of residents. As an important part of community services, intelligent medical care is even more essential. Faced with the situation of the increasing number of chronic diseases caused by the poor living habits of contemporary community residents and the untimely physical examination, we establish the BP neural network(BPNN) prediction model for hypertension. In view of the problem of single value missing caused by incomplete data collection and data loss in the intelligent community, we propose an algorithm based on BPNN prediction interpolation, which predicts the interpolation missing value information with known health information, and then predicts the risk of hypertension of residents with the completed data. That is, a method that can predict hypertension in samples with missing values. Experimental results show that the BPNN interpolation method has higher accuracy than the traditional interpolation method, with error of 5.3%,and the data is applied to high blood pressure after interpolation prediction effect is better, with accuracy of 93%. The model can predict hypertension for incomplete data samples, provide forecast results before the residents medical examination and save medical resources, which provide certain medical security services for community residents.