Distributed Mentor Project 2002

by Tu Tran

About Me

About My Mentor, Prof. Jennifer Mankoff

Project Journal

User Study Lessons

Readings on HCI and Nutrition

Final Report



Project Description

The prevalence of obesity (defined as a body mass index >= 30 kg/m ) increased from 12.0% in 1991 to 17.9% in 1998 in the United States, according to the study The Spread of the Obesity Epidemic in the United States, 1991-1998.  This is a major public health concern since obesity, according to reports cited by the study, is strongly associated with several chronic diseases such as cardiovascular diseases and diabetes.  In a Centers for Disease Control and Prevention (CDC) on-line article regarding the study, Jeffrey P. Koplan, director of the CDC and one of the co-authors of the study, stated, "Overweight and physical inactivity account for more than 300,000 premature deaths each year in the United States, second only to tobacco-related deaths." 

A healthier diet, in addition to more exercise, can help abate weight gain and obesity.  The goal of the Nutrition Project, led by Professor Jennifer Mankoff and initiated by a team of Berkeley students, is to create a ubiquitous computing device that gives nutrition advice based on scanned grocery receipts.  By allowing users to scan in their grocery receipts, the device tries to impact the lives of users as little as possible (low-fidelity sensing) and at the same time suggest food alternatives which contain less fat and more overall daily nutrients than the foods they have purchased.    

The technology behind the project consists of an optical character recognition (OCR) program, a database, and an inferencing system.  The receipts are scanned in and passed through the OCR program, and then the data from the program is passed to a MySQL database.  The database stores user profiles, historical information about user shopping habits, and data about the nutritional contents of foods.  The data about the nutritional contents of foods is from the USDA Nutrient Database for Standard Reference, Release 14.  The inferencing system  compares the scanned food products to the recommended nutrient consumption, as recommended by the USDA Food and Nutrition Information Center.  It then outputs a list of alternative foods, within the same food groups, that contain more recommended nutrients and less fat than the purchased foods.  

For more detailed information on the project and technology, see the project's paper "Using Low-Cost Sensing to Support Nutritional Awareness" which has been accepted to the UbiComp 2002 conference.

My primary project this summer involved implementing a long-term user survey.  The goal of the  survey is to track the reported and purchased consumption habits of individuals over a several month period, through collected food frequency questionnaires and receipts.  This data will then be used to improve the inferencing system and to experiment with different data aggregation periods to determine what is optimal for each food group. 



CRA-W coordinated and sponsored my research this summer.

I took part in the weekly GUIR meetings where research projects and issues  were presented and discussed.

UbiComp 2002

A paper based on the project before my participation was accepted at the UbiComp 2002 conference.


The views and opinions expressed in these personal pages are the sole responsibility of the author and are not official statements of Mills College, nor is the College responsible for their content. (see full disclaimer)