基于结合算法的血压预测模型的探究与实现

ata mining, as a means of extracting useful information from data, can help us analysis existed medical data and provide scientific decision-making for diagnosis.Recently, in the research of blood pressure estimation, one of the most prominent methods is to use the Pulse Transit Time(PTT).It is proven that PTT has a strong correlation with the blood pressure, however, this method ignores that PTT is highly dependent to each individual's physiological properties, thus estimation results need to be revised for each individual. An improved method which has been proposed by Mohamad Kachuee utilizes multiple features of ECG and PPG diagrams to establish a regression model to reduce the error of Pulse Transit Time.This paper implements three BP estimation models combining linear regression, CART regression and SVM regression based on combination algorithm.Each model minimizes the mean prediction error by adjusting the weight of the internal regression models. Finally, we calculate the mean absolute error, standard deviation and running time, it is concluded that the BP estimation model combing CART and SVM regression model performs best from the three aspects of accuray, stability and efficiency.