Steady State Detection for the Context Aware Evaluation of Vital Signs

In the context of Ambient Assisted Living, the measurement and analysis of vital signs in the context of executed activities take a key role in knowledge processing. Regarding the typical attributes of time series covering health and metabolism parameters like respiratory rate, heart beat or other dynamic values, this article proposes an iterative multi stage approach for data handling in accordance to the Knowledge Discovery in Databases - KDD approach. In order to increase the significance of the raw data by preprocessing, procedures for the treatment of missing values, artifacts, outliers and the removal of noise components have been implemented. The consecutive step of feature extraction, the identification of segments, steady states and transients in the vital sign series which are directly linked to certain activities or activity transitions, closes the proposed data analysis process. Finally, experiments including real measurements with the subsequent discussion of the measurement results provide an outlook to the capabilities of our approach and grant information about open issues and the steps in research.