Personalization goals for run-time adaptation of IoT-based assistance applications for the elderly

Context: the increasing demand for Ambient Assisted Living (AAL) applications has led to the need for personalized assistive tasks that can adapt to individual users' needs. Objectives: we aim to balance design-time personalization with techniques of run-time adaptation for designing and executing assistive AAL applications, personalized to both users' specific needs and environmental conditions. Methods: we propose a personalization process based on: (1) representing assistive tasks as workflows initially defined at a high level of abstraction that specifies their functional components, (2) providing an instrument for specifying how to customize these workflows for individual users, and (3) a supporting architecture that enables the run-time transformation of high-level specifications into executable workflows. Results: our empirical evaluation demonstrates that the proposed personalization goals effectively support designers in creating adaptable workflows, showing improved quality scores in personalization compared to traditional BPMN practices, without increasing design effort. Performance analysis also shows the feasibility of our run-time adaptation approach with linear scaling as the number of personalization goals increases. Conclusion: a personalization process for modelling personalizable workflows may be a flexible instrument for designers to conceive assistive applications that are automatically adapted to individual users' needs at run-time, allowing for balancing the benefits of design-time and run-time personalization techniques.