A wavelet-based approach for time series pattern detection and events prediction applied to telemonitoring data

This work aims the development of a predictive strategy able to estimate future events with relevant impact in the cardiovascular status. Based on wavelet transform, a new time series similarity metric is introduced, which is capable to detect a pre-defined pattern in time series data. In addition, a methodology combining a wavelet scheme with state space multi-models is proposed to achieve the prediction of future signal values. Blood pressure signals, collected by a telemonitoring platform (TEN-HMS), are used to detect the occurrence of future hypertension events.