Testing for mobile E-health interventions

Evaluating interaction that spans several weeks, between a virtual coach and a human patient on a mobile behavior intervention is virtually impossible. Traditional laboratory-based user testing is too tedious and expensive, and allows only the testing of a limited amount of different scenarios. Ideally, the test results of a mobile behavior intervention application cover a clear understanding of: (i) different user models, (ii) adaptive behavior of the coach, and (iii) different levels abstraction from multiple stakeholder viewpoints. Towards this goal, we address challenges posed by the domain and devised an automated testing environment using a simulator that mimics the behavior of a patient during the process of behavior change based on scenarios. The simulation results, i.e. context sensitive user interaction (on the application) at different phases and stages of a six-week insomnia therapy, can be recorded and presented to stakeholders for the application quality assessment and analysis.