JMIR Publications

ABSTRACT

Background:

A growing body of evidence shows that financial incentives can effectively reinforce individuals' positive behavioral change and improve compliance with health intervention programs. A critical factor for the design of incentive-based interventions would be setting a proper incentive magnitude. However, it is highly challenging to determine such magnitudes because the effects of incentive magnitude highly depend on personal attitudes and contexts.

Objective:

This study aims to illustrate an adaptive micro-contingency management (AMCM), in which users' micro-behavioral changes are rewarded with particular incentive amounts that are adaptively estimated based on individuals' prior behaviors on different incentives and contexts, and to describe lessons learned from a feasibility study.

Methods:

We implemented a mobile health intervention app for preventing prolonged sedentary lifestyles, in which participants received a behavioral suggestion (i.e., an active break mission) with a particular incentive that was offered after adherence to the suggestion when they uninterruptedly sat for 50 minutes. Participants were distributed into either the fixed (i.e., rewarding positive behavior with the same incentive) or the adaptive incentive group (i.e., rewarding positive behavior with varying incentives estimated by the AMCM), and the intervention lasted for three weeks.

Results:

We recruited forty-one participants (15 females; fixed incentive group: 20 participants; adaptive incentive group: 21 participants) whose mean age was 24.0 (SD: 3.8; range: 19-34). Mission success rates did not show statistically significant differences by group (0.66 and 0.61 for the fixed and adaptive groups, respectively; P=.541). The follow-up analysis on the adaptive incentive group revealed the influence of incentive magnitudes on mission success was not statistically significant (OR=0.98; P=.184). Based on the qualitative interview, such results were possibly due to participants having had sufficient intrinsic motivation and less sensitivity to incentive magnitudes.

Conclusions:

While our AMCM did not show significant effects on users' behavior occurrence likelihood, this study configures a pioneer work toward adaptively estimating incentives by user behaviors and contexts through leveraging mobile sensing and machine learning. We hope that this study inspires researchers in developing various incentive-based interventions.