Improved deep belief network (DBN)-based cardiovascular disease (CVD) predicting model

Abstract

The invention discloses an improved deep belief network (DBN)-based cardiovascular disease (CVD) predicting model. The model is based on DBN; the feature abstract expression is carried out layer by layer by utilizing multi-layer network architecture; optimal network parameters obtained through training are used for initializing neural network; in the meantime, the network depth is autonomously determined by utilizing reconstruction errors; the improved DBN-based predicting model is constructed through combining unsupervised training and supervised tuning; and therefore, the stability is guaranteed while the accuracy of the model prediction is increased at the same time. By means of the model disclosed by the invention, the problem that the prediction accuracy of a traditional predicting model is reduced under the condition of multi-classification and non-linear complex factors is solved; and the problem that the variance of prediction results is increased due to the randomness of initial parameters of shallow neural network is also solved at the same time.

Description

technical field [0001] The invention relates to the field of prediction models, in particular to a cardiovascular disease prediction model based on an improved deep belief network. Background technique [0002] At present, cardiovascular disease prediction models are mainly divided into: one is the traditional prediction model based on probability. The inferential prediction method based on the trend reasoning based on the epidemic process and characteristics of the disease, relying on the knowledge level and subjective experience of experts on cardiovascular disease (CVD) to make qualitative predictions; the mathematical prediction based on the cross-sectional data of large follow-up cohorts such as disease risk factors and incidence Methods, such as time series prediction models, regression prediction models, etc., establish mathematical probability models for regularly changing data, and quantitatively mine the proportional relationship between pathogenic factors, which r...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to overcome the deficiencies in the prior art, provide a kind of cardiovascular disease prediction

Images