Fingerprints: detecting meaningful moments for mobile health intervention

Abstract

Personalized and contextual interventions are promising techniques for mobile persuasive technologies in mobile health. In this paper, we propose the "fingerprints" technique to analyze the users' daily behavior patterns to find the meaningful moments to better support mobile persuasive technologies, especially mobile health interventions. We assume that for many persons, their behaviors have patterns and can be detected through the sensor data from smartphones. We develop a three-step interactive machine learning workflow to describe the concept and approach of the "fingerprints" technique. By this we aim to implement a practical and light-weight mobile intervention system without burdening the users with manual logging. In our feasibility study, we show results that provide first insights into the design of the "fingerprints" technique.