Self-Quantification Systems to Support Physical Activity: From Theory to Implementation Principles

Abstract

Since the emergence of the quantified self movement, users aim at health behavior change, but only those who are sufficiently motivated and competent with the tools will succeed. Our literature review shows that theoretical models for quantified self exist but they are too abstract to guide the design of effective user support systems. Here, we propose principles linking theory and implementation to arrive at a hierarchical model for an adaptable and personalized self-quantification system for physical activity support. We show that such a modeling approach should include a multi-factors user model (activity, context, personality, motivation), a hierarchy of multiple time scales (week, day, hour), and a multi-criteria decision analysis (user activity preference, user measured activity, external parameters). This theoretical groundwork, which should facilitate the design of more effective solutions, has now to be validated by further empirical research.

Keywords: behavior change; health; model; persuasive design; physical activity; quantified self; support system; user-centered design.

Figure 1

Adapted from Li et al.’s Stage-Based Model of Personal Informatics Systems: this shows the progression of a person toward behavior change through the different stages of a self-quantification experience with its iterative nature and its barriers.

Figure 2

Adapted from Epstein et al.’s lived informatics model of personal informatics: this model is based on Li et al.’s model and highlights the essential fluidity and iteration of a self-quantification process. It is not specifically oriented towards behavior change, though.

Figure 3

Adapted from Vizer et al.’s conceptual model of shared health informatics (CoMSHI): also based on the stage-based model, the CoMSHI enhances the fluidity of the process by facilitating transitions between stages. It reflects the need for context raised by previous research as well.

Figure 4

Adaptive System for Physical Activity: Support Phase—Daily Time Scale. After the initial learning phase, we know the user’s activity patterns, as well as physical health, personality, and context that compose the user profile. Hence, we are able to determine an optimal challenge point for the current user day based on his/her profile before monitoring the progress in separate intraday loops.

Figure 5

Adaptive System for Physical Activity: Support Phase-Intraday Sub-Goal Time Scale. An ideal sub-goal (3000 steps halfway through the day for example) is determined according to the objective of the day (e.g., 6000 steps). A control loop is run hourly to monitor user physical activity level and to evaluate if she is making good progress toward the sub-goal. If self-management (left loop) is not sufficient, the system can intervene to propose the user a personalized physical activity adapted to the current context (right inner loop), or move to the next sub-goal in case of failure (right outer loop).

Figure 6

Personalized and Adapted Activities Suggestions Process: this figure details how a self-quantification system for physical activity support should rely on a user preference model of activities, (1) before filtering it with monitored contextual elements (2) in order to produce its recommendations (3). A personalized list of context-sensitive activities is proposed to the user from which they can choose.