Real-time modelling of heart rate response during exercise using a novel constrained parameter estimation method.

This paper is devoted to the problem of real-time heart rate (HR) response modelling during treadmill exercise. A novel recursive constrained parameter estimation method is developed which in contrast to the conventional parameter estimation schemes (e.g. recursive least squares (RLS) method) can avoid the occurrence of the so-called blowup phenomena. By incorporation of a weighting upon 1) parameter variation relative to a priori HR response knowledge, 2) one-step parameter variation, into the objective function, an estimation scheme is obtained that in the absence of exciting data can avoid blowup. The proposed estimation scheme were experimentally verified using eight healthy male subjects and the results demonstrated that the designed scheme is able to identify the HR response of the exercising subjects in a real-time manner. As HR response is naturally a time-varying dynamics, the proposed online modelling method is of importance for the HR regulation during exercises, using a feedback controller with a desirable level of performance.