Measuring Heart Rate During Physical Exercise by Subspace Decomposition and Kalman Smoothing

Monitoring of health parameters during physical exercise is an important aspect of both sports and rehabilitation medicine. Photoplethysmography (PPG) is routinely employed for low-cost heart rate (HR) measurement; however, monitoring during physical exercise is made difficult by the presence of motion artifacts. In this paper, we present an approach that combines denoising by subspace decomposition and Fourier-based HR measurement, and finally, smoothing and tracking by a Kalman filter. Using publicly available real-life PPG traces, we demonstrate accuracy and performance by an extensive set of experimental results, comparing them with similar algorithms proposed in the literature.