On the utility of increasing the number of matches in computing sample entropy over short cardiovascular variability series

Strategies applied to artificially increase the number of matches in conditional entropy (CE) estimation were tested against the traditional sample entropy (SampEn) definition. The comparison was carried out in a practical context of cardiovascular control complexity assessment via heart period and systolic arterial pressure variability analysis in Parkinson disease patients and age- and gender-matched healthy subjects at rest in supine position and during postural challenge via head-up tilt at 75°. The use of these strategies did not produce any practical advantage compared to traditional SampEn definition, likely because the reduction of the variance of the CE estimates was accompanied by the decrease of its mean due to the artificial increase of regularity. Also the possible blurring of nonlinear patterns and mixing of oscillation phases might have contributed to this disappointing result.