Effects of Sampling Frequency and Data Length on the Central Aortic Waveform Reconstruction

摘要: Non-invasive CAP (central aortic pressure) reconstruction is mostly based on the auto regressive eXogenous (ARX) model or Fourier transform in transfer function method, without considering the factors such as sampling frequency and data length. Based on the ARX model and the Fourier transform, the error of reconstruction CAP was analyzed for studing the effects of sampling frequency and data length on the CAP. The results show that when sampling frequency is 100Hz and data length is greater than 3s, CAP can be better reconstructed by ARX model (RMSE:(306.6±80.0) Pa; FIT: 89%). The algorithm based on the Fourier transform is insensitive to sampling frequency. When data length is set to 6s, the reconstructed CAP has a better performance (RMSE:(493.3±320.0) Pa, FIT: 84%).