Mobile Health Behavior Tracking: Health Effects of Tracking Consistency and Its Prediction
42 Pages Posted: 13 Apr 2020 Last revised: 1 May 2020
Date Written: April 6, 2020
New technologies aimed at nudging large numbers of individuals towards healthier behavior (e.g., fitness trackers, wearables, A.I.-based health coaching) increasingly focus on allowing users to track their own health goal-directed behavior on an ongoing basis, in the hope to boost their progress. However, little is known about (1) whether or not consistent personal health behavior tracking actually yields noticeable health benefits (e.g., weight loss) in the long run and, if so, (2) what factors predict consistent tracking activity over time. Using data from a popular mobile fitness app, we use a novel machine learning method for flexible instrumental variable discovery to show that greater consistency in calorie tracking (i.e., greater frequency and continuity) leads to greater weight loss in the long-term. The importance of tracking in itself then motivates our predictive analytics, where we assess the importance of progress-based (e.g., past weight loss, staying within one’s calorie budget), behavior-based (e.g., last period’s exercise and food calories, past tracking behavior), and demography-based (e.g., age, gender, initial weight, initial distance to goal weight) features for predicting consistent tracking. This predictive analysis reveals that while behavior-based features are among the most important predictors, progress-based predictors are also important (e.g., past calories over/under budget).
Keywords: Behavioral Analytics; Digital and Mobile Health Marketing; Predictive Analytics; Self-Control and Motivation
Suggested Citation: Suggested Citation