Automated Personalized Health Feedback
Main investigators – Tanzeem Choudhury, Hane Aung, Mi Zhang and Mashfiqui Rabbi
Past investigators and collaborators – Bonnie Spring, Angela Pfammatter, Erin Carroll, Steve Voida, Jean Costa, Max Schachere, Chantelle Farmer
Modern fitness devices and apps can keep track of fine-grained personal health information. However, none of them does anything significant with the data to promote healthier lifestyle. MyBehavior changes all of that and is the first mobile app that provides personalized health suggestions automatically. MyBehavior’s learns its user’s lifestyle and suggests small changes that users can actually do. MyBehavior is like Netflix or Pandora for your health behavior. It is different and smarter than mere step-counting, or one-size-fits-all suggestions that healthcare professionals or apps give now-a-days.
Under the hood, MyBehavior uses state-of-the-art artificial intelligence technique: an online multi-armed bandit model automatically generates context-sensitive and personalized activity and food suggestions based on calorie count and the user’s physical activity and food habits. MyBehavior continually adapts its suggestions by exploiting the most frequent healthy behaviors while sometimes exploring non-frequent behaviors in order to maximize the user’s chances of reaching her health goal (e.g., weight loss). Also, MyBehavior takes user privacy seriously. It creates all the suggestions inside the phone without exporting any data to the cloud.
(Click on the image to enlarge)
Figure: (a) Personalized physical activity suggestions generated by MyBehavior (b) progress of the first suggestion in ‘a’ over time (c) location of first suggestion (d) progress of another suggestion overtime (e) personalized food suggestions based on what the user has eaten before.
- Mashfiqui Rabbi, Min Hane Aung, Mi Zhang and Tanzeem Choudhury. MyBehavior: Automated Personalized Health Feedback from User Behavior and Preference using Smartphones. The 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing (Ubicomp 2015).
- Mashfiqui Rabbi, Angela Pfammatter, Mi Zhang, Bonnie Spring, and Tanzeem Choudhury. Automated Personalized Feedback for Physical Activity and Dietary Behavior Change With Mobile Phones: A Randomized Controlled Trial on Adults. JMIR mHealth uHealth 2015;3(2):e42
- Mashfiqui Rabbi, Jean Costa, Fabian Okeke, Max Schachere, Mi Zhang, and Tanzeem Choudhury. An Intelligent Crowd-worker Selection Approach for Reliable Content Labeling of Food Images. The Proceedings of Wireless Health 2015.
- MIT Technology review, A Health-Tracking App You Might Actually Stick With