Population-based Reconstruction Of Central Aortic Pulse Wave Using ARX Model

At present cardiovascular disease is a great threat to human health.From human pulse wave one can obtain a lot of information reflecting the state of the cardiovascular system.Traditional peripheral pulse wave studies have limited the current access to more accurate clinical information.Therefore,the measurement of central arterial pulse wave is of great interest.The current approach to measure central arterial pulse wave using interventional catheters in clinical practice is the "gold standard",however this approach results in great pain to the patients and great economic stress to their family.This invasive method is usually used during cardiac surgery,but not merely as a method in the measurement of central arterial pulse wave.The representative non-invasive detection systems are SphygmoCor(AtCor,Austrilia)pulse wave analyzer and Omron's Omron HEM-9000 AI pulse measuring instrument etc.However,these two non-invasive detection products employ the general transfer function method and they are suspected for being lake of accuracy.This thesis proposed a population-based method to reconstruct central arterial pulse wave using an ARX model.Firstly,an ARX model was built based on 25 data sets of central arterial and brachial pulse waves.The model order was determined by AIC and FPE criterion.The model parameters were calculated using Least Squares method.The generalized transfer function was obtained by averaging all transfer functions derived.Secondly,the transfer functions of all subjects were grouped according to sex,age,BMI,and K-value.Finally,the accuracy of the sub-grouped transfer function was tested and compared with that of the generalized transfer function.During the validation,the central arterial pulse wave was estimated by the transfer function,and based which several feature parameters were derived.Comparison was presented between the TF-estimated parameters and the measured ones.T-test was employed to verify that there is no significant difference between the TF-estimated parameters and the measured ones.And the consistency between the estimated parameters and the measured parameters was validated by the Bland-Altman method.The correlation between the TF-estimated parameters and the measured ones was also calculated.In conclusion,the validation in 25 subjects suggests that the method proposed in this thesis indeed improves the accuracy of the transfer function method in estimating central arterial pulse wave.Among the group methodes,the one based on age and BMI has the best performance.