A Resonance Cardiorespiratory Coupling Model: Learning and Validation from Human Heart-Rate and Respiration Data

The resonance phenomenon in the cardiorespiratory coupling system has long been discovered and its application is beneficial to human health. Among cardiovascular models that study this phenomenon, the DeBoer's model is an extensively studied physiological model which describes the relationship of the heart rate (HR), the blood pressure (BP) and the respiration. However, the Deboer's model does not have good prediction performance in fixed rate breathing. In this paper, the DeBoer's model is modified by supplementing the neural effect of respiration on HR, which is a basic mechanism. We identify parameters of the modified model and the DeBoer model by LS method based on training sets of 14 subjects in our experiment, and validate their prediction performances based on test sets. The result shows that the prediction performance of the modified model is much better than the DeBoer's model: the phase accuracy and time-series accuracy of HR rises from 69.04% to 86.37% and from 93.3% to 95.25% respectively. It is validated that our modification to the model is effective.