A robust adaptive estimator of rate for cardiac arrhythmia detection

A robust adaptive algorithm for rate estimation is proposed for use with implantable cardioversion/defibrillation devices. The algorithm consists of a two-stage processor employing outlier detection and removal followed by an adaptive trimmed mean filtering operation. Outliers are detected by comparing raw values with the median absolute deviation (MAD). The residual data are subjected to a trimmed mean filter, where the trimming parameter is adapted according to a local measure of tail behavior. The algorithm is tested using both simulated data from known distributions and experimental data. The authors´ result indicate significant reductions in mean-squared estimation error when compared to averaging and other simple estimators, and a performance that is close to the minimum mean-squared error corresponding to the fixed optimal estimator, but without prior knowledge of the data distribution