Postural time-series analysis of elderly and control subjects using entropy

The present paper applies two methods commonly used to calculate entropy to postural time-series data of elderly and control subjects. Approximate entropy provided a biased estimate of regularity for small window lengths. In contrast, sample entropy was able to identify postural differences between elderly and control subjects. Elderly subjects showed less regularity in terms of anteroposterior displacement than did control subjects. However, the observed values of entropy were very small, as were the significant differences observed. Additional work is required to address concerns related to the time series length and the method used to calculate entropy.