Are strategies favoring pattern matching a viable way to improve complexity estimation based on sample entropy?