Visualizing Uncertainty in Multi-Source Mental Health Data

Abstract

Self-tracking technologies, especially those facilitating support from social systems, are becoming more common for treating serious mental illnesses in both clinical and informal contexts. A recently proposed feature is co-tracking, where data is gathered not only from the perspective of the user managing their condition, but also from their close contacts. The proposed system therefore supports multiple perspectives (data streams) about the same variable of interest (i.e., an individual's mood). However, the subjective and reciprocal nature of mental health data gives rise to challenges in visualizing uncertainty that must be addressed before clinical use. Here, we create an application-specific typology of uncertainty for visualizing multi-source mental health data, and propose design solutions to communicate this uncertainty. Via a case study of mood tracking with bipolar disorder, we present an interactive visualization prototype for understanding dynamic mood states in close relationships, moving toward a real-world implementation of a co-tracking informatics system.