AAAI Publications, 2018 AAAI Spring Symposium Series

Font Size: 
A Dynamic Learning Model for a Better Personalized Healthcare Using Mobile Health Tools
Amy Wenxuan Ding

Last modified: 2018-03-15


Scientists estimate nearly half of the world's adult population will be overweight or obese by 2030. Widely used mobile devices can provide inexpensive tools to reinforce self-monitoring of weight management behaviors and have a great potential in obesity treatment. However, their effectiveness depends on whether users actively responds to suggestions or health interventions displayed. This paper proposes a novel theory-based dynamic learning model to examine how a user’s unobserved mind states of activated engagement in weight loss affect her weight management activities. Based on a mobile health app dataset, we find that there exist two mind states (activated vs. inactivated) among the app users. Users in the activated state of weight loss engagement significantly increase their daily steps taken by 57.82% compared to those in the inactivated state when following the health interventions in the app. Further, a simple home-screen reminder of checking the health suggestions in the app targeting inactivated-state users will increase their probabilities and time duration of moving into the activated state by 29% and 38.9%, respectively. As a result, user mind state-based personalized healthcare interventions in the mobile app are shown to be quite effective.


mobile health; artificial intelligence; mind states; dynamic learning

Full Text: PDF