Online Change Detection for Timely Solicitation of User Interaction

The accurate detection of changes has the potential to form a fundamental component of systems which autonomously solicit user interaction based on transitions within an input stream, for example accelerometry data obtained from a mobile device. This solicited interaction may be utilized for diverse scenarios such as responding to changes in a patient’s vital signs within a medical domain or requesting activity labels for generating real-world labelled datasets. Within this paper a change detection algorithm is presented which does not require knowledge of the underlying distributions, can run in online scenarios and considers multivariate datastreams. Results are presented demonstrating practicable potential with 99.81% accuracy and 60% precision for real-world accelerometry data.