CinC challenge 2002 undertaken by non-stationary and fractal techniques

By using a local recursive least squares approach, named the αβ filter, to improve Detrended Fluctuation Analysis, we addressed the identification of real and synthetic Heart Rate Variability (HRV) data proposed as part of the Computers in Cardiology Challenge 2002. This approach provided power-law patterns that were used to correctly classify the majority (46/50) of the Challenge dataset. In addition, we show that this approach revealed patterns in real HRV data that do not necessarily follow a uniform power-law. Consequently, this paper promotes familiarisation with scaling information that has not been reported previously for healthy HRV data.