Matlab tool applied to the trend prediction of physiological time series

This work describes a Matlab tool implemented to address the trend prediction of physiological time series. It is composed of two main components: time series similarity analysis and wavelet decomposition prediction modules. This tool permits to adjust a set of parameters and configurations that can be used by users/professionals in an active training process. Moreover, a set of common prediction methods is available, enabling the comparison of the corresponding performances. The scheme was implemented using data collected by means of TEN-HMS and myHeart tele-monitoring studies. In particular, the estimation of future trends of daily blood pressure, heart rate and weight were addressed. Additionally, for patients whose blood pressure values are in a critical range, the assessment of the hypertension is also estimated.