基于脉搏波的无创连续血压监测模型簇研究

Research on the non-invasive continuous blood pressure monitoring models cluster based on pulse wave

脉搏波中蕴含丰富的血压信息,且检测简便,适用于血压的无创连续监测。利用脉搏波信号,结合心电信号,提取其波形特征,引入个体特征,基于误差逆传播神经网络构建关于收缩压、舒张压的监测模型。在模型构建过程中,使用相关性分析、平均影响值法减少特征冗余;利用自组织特征映射神经网络完成具有近似属性的样本的分类;使用多种群遗传算法确定网络的初始权重、阈值,分类别构建模型,形成血压监测模型簇;最后再利用多种群遗传算法进行个体参数的优化,得到最终的个体血压监测模型。结果显示,该模型的血压预测值与实测值具有极强的相关性;模型估计误差满足美国医疗仪器促进协会标准(5±8 mmHg)以及英国高血压协会标准的A级标准。该模型在一定程度上增加了血压监测过程中的模型自校正能力,有望应用于长时无创连续血压监测设备当中。

The pulse wave contains abundant information of blood pressure.In addition,it is convenient to measure the pulse wave.Therefore,it is appropriate to implement non-invasive continuous monitoring of blood pressure.In this study,the pulse wave and the electrocardiogram signal are both utilized to extract waveform information.The individual characteristics are introduced.The systolic and diastolic blood pressure estimation models are formulated by using the error back-propagation(BP)neural network.During formulating the model,the correlation analysis and the mean impact value method are employed to reduce feature redundancy.And the self-organizing feature mapping neural network is used to achieve the classification of approximate attribute samples.The multiple population genetic algorithm is applied to determine the initial weights and thresholds of the network.And the blood pressure monitoring models are established by classification to form a blood pressure monitoring models cluster.Finally,the multiple population genetic algorithm is utilized to optimize the personalized parameters to realize the final individual continuous blood pressure monitoring model.Experimental results show that the predicted values of models have strong correlation with the measurement values of the electronic sphygmomanometer.The estimation error of the models meets the requirement of both the Association for the Advancement of Medical Instrumentation criteria(5±8 mmHg)and the Grade A of British Hypertension Society criteria.The proposed models increase the self-calibration ability in the process of blood pressure monitoring to a certain degree.It is expected to be applied in the long-term non-invasive continuous blood pressure monitoring equipment.