Análise de Desempenho da Entropia Aproximada (ApEn) na Análise da Variabilidade da Frequência Cardíaca (VFC)

This work evaluates the performance of a nonli-near indicator in front of noise in the analysis of Heart Rate Variability (HRV) due to the presence of artifacts. For this purpose, we select approximate entropy (ApEn). Firstly, we use artificial RR time series and compare the error differences between the computed values before and after inserting ran-dom false-positive and false-negative beats. Afterwards, we use ambulatory recordings of 18 healthy subjects from the MIT-BIH database and compare the error difference between the computed values mantaining and excluding artifacts. We have found that the average errors, both in artificial RR time series and ambulatory recordings, vary inversely proportional to the metric value of ApEn for each time series without artifacts.