Nonparametric dynamical model of cardiorespiratory responses at the onset and offset of treadmill exercises

Abstract

This paper applies a nonparametric modelling method with kernel-based regularization to estimate the carbon dioxide production during jogging exercises. The kernel selection and regularization strategies have been discussed; several commonly used kernels are compared regarding the goodness-of-fit, sensitivity, and stability. Based on that, the most appropriate kernel is then selected for the construction of the regularization term. Both the onset and offset of the jogging exercises are investigated. We compare the identified nonparametric models, which include both impulse response models and step response models for the two periods, as well as the relationship between oxygen consumption and carbon dioxide production. The result statistically indicates that the steady-state gain of the carbon dioxide production in the onset of exercise is bigger than that in the offset while the response time of both onset and offset are similar. Compared with oxygen consumption, the response speed of carbon dioxide production is slightly slower in both onset and offset period while its steady-state gains are similar for both periods. The effectiveness of the kernel-based method for the dynamic modelling of cardiorespiratory response to exercise is also well demonstrated. Graphical Abstract Comparison between VO and VCO during onset and offset of exercise.

Oxygen Uptake Response to Switching Stairs Exercise by Non-parametric Modeling

“…In previous research, the first-order system was proposed for V O 2 estimation during the single onset or offset period experiment [11] [33] [34]. However, the first-order system is not suitable when the input and output of the system become complicated as in our experiment.…”

Nonparametric Model Prediction for Intelligent Regulation of Human Cardiorespiratory System to Prescribed Exercise Medicine

“…In the numerical simulation part for parameter selecting, we acquired the β value under the assumption that the system is a first-order system based on the previous study [20]. Our previous study [9] also revealed that in the single ascending or descending period, the system is close to a first-order system, but there are still exceptions because of the individual difference. However, in this research, the physiological information contains two periods of ascending (0 − 3 − 8km/h) or a full period with ascending and descending (3 − 8 − 3km/h).…”

“…Based on our previous work [9], the Stable Spline (SS) kernel which is shown in Eq. 7, demonstrates a better performance than the other kernels based on the aspects of accuracy, sensitivity, stability and smoothness.…”

“…Two approaches are compared in the selection of the kernel parameter in this methodology. One approach is the numerical simulation which has been proposed in our previous paper [9], and the other approach is selecting the parameter of the kernel in the regularization term based on identification results when using data from the stairs experiment.…”