基于主成分分析的不受睡眠姿势影响的呼吸容积动态监测方法

Respiration Volume Dynamic Monitoring Method of Not Being Affected by Sleep Postures Based on Principal Component Analysis

目的探究基于主成分分析(PCA)的不受睡眠姿势影响的呼吸容积动态监测方法。方法利用2014年6月采集到的10名健康男性大学生平静呼吸时在仰卧、左侧卧和右侧卧位下的呼吸流速(PNT)信号及胸腔阻抗(IPTH)、左肺阻抗(IPL)和右肺阻抗(IPR)信号,通过对PNT信号进行积分运算得到呼吸容积(VPNT)信号作为金标准,使用基于PCA的数据融合算法从左肺和右肺的呼吸信号中获取融合的容积信号(VIPLR)。结果VIPLR与VPNT的相关系数(r)高达0.9501,且与经典的胸腔阻抗容积(VIPTH)比较,容积平均绝对误差(MAE)下降了25%。结论基于PCA的睡眠呼吸容积评估方法是一种可信度较高的睡眠呼吸容积动态监测新方法,能够最大限度地克服睡眠姿势变换对监测结果的影响。

Objective The respiration volume dynamic monitoring method of not being affected by sleep postures based on principal component analysis(PCA)was explored in this study.Methods The signals of pneumotachograph(PNT),the thorax impedance(IPTH),the left lung impedance(IPL)and the right lung impedance(IPR)of 10 healthy male college students were collected in June 2014 under the condition of eupnea in three gestures i.e.the dorsal decubitus,left lateral decubitus and right lateral decubitus.The respiration volume signals(defined as VPNT)was obtained after the PNT signals were integrated and VPNT was regarded as the gold standard.The fused respiratory volume(VIPLR)signals were obtained through calculating the left and right lung respiratory impedance signals by the data fusion algorithm based on(PCA).Results The correlation coefficient(r)between VIPLR and VPNT was up to 0.9501,and compared with the classical thorax impedance volume(VIPTH),the mean absolute error(MAE)of volume was reduced by 25%.Conclusion Sleep respiration volume evaluation method based on PCA is a new and high-reliability method for sleep respiration volume dynamic monitoring,and can reduce the effects of posture changes on the monitoring results to the greatest extent.