An exercise-driven heart rate statistical process model

Relationship between heart rate and exercise is useful for health diagnosis and monitoring. A statistical process model is proposed to model relationship between heart rate and activity. Multiple models are built and each of them is corresponding different phase of heart-rate change due to activity, such as drift and recovery phases. Subject-dependant parameters of each model are estimated using LMS method. An integral model for full heart rate dynamic are built by further fusing those model based on model probability. The variant models for both heart predication given activity intensity and real time heart rate filtering are also presented The multi-phase model is more consistent with physiological process and the statistical nature of it makes it more robust. And more importantly, it is the first model which also describes the variability of heart rate changes with exercise, besides kinetics.