基于改进深度信念网络的心血管疾病预测研究

Research on cardiovascular disease prediction model based on improved depth trust network

心血管疾病传统预测模型准确率较低,基于浅层神经网络的模型预测结果方差较大。为此提出一种基于改进深度信念网络的心血管疾病预测模型,利用重构误差,自主确定网络深度,结合无监督训练和有监督调优,在提高模型预测准确率的同时保证稳定性。对UCI数据库中的statlog (heart)和heart disease database独立进行30次实验,结果显示预测准确率的均值分别为91. 26%、89. 78%,预测准确率的方差分别为5. 78、4. 46。

The prediction accuracy of the traditional cardiovas cular disease model is low. The results of model prediction based on shallow neural network are too large. This paper proposed a cardiovascular disease prediction model based on improved depth trust network. It used the reconstruction error,autonomously determined the depth of the network. Combined unsupervised training and supervised optimization,it could improve the prediction accuracy and stability of the model. Using the prediction model based on depth learning mentioned,it tested the statlog（ heart） and heart disease dataset database 30 times independently in experiments. The results show that the accuracy rate is 91. 26% and 89. 78% respectively,and the variance is 5. 78 and 4. 46 respectively.