The paper provides a brief discussion on the need of providing context-aware systems for supporting ageing people in their every day life. The authors describe the initial model of a layered and extensible context architecture that provides self-management and self-configuration capabilities. In particular they propose to include prediction as a means to address context modeling through pattern recognition. Currently the proposed architecture has not been implemented and only a sample application for monitoring user behavior and for visualizing deviations from the learned patterns is described.