Related Experiment Video
Updated: Nov 25, 2025
06:21
Concept Development and Use of an Automated Food Intake and Eating Behavior Assessment Method
Published on: February 19, 2021
6.1K
A Real-Time Eating Detection System for Capturing Eating Moments and Triggering Ecological Momentary Assessments to
Mehrab Bin Morshed1, Samruddhi Shreeram Kulkarni1, Richard Li2
1Georgia Institute of Technology, Atlanta, GA, United States.
JMIR Mhealth and Uhealth
|December 18, 2020
Summary
This study developed a smartwatch system to detect eating episodes and collect contextual data using ecological momentary assessments (EMAs). The system accurately identified meals, revealing that most eating occurs with distractions and often alone, impacting well-being.
Area of Science:
- Behavioral Science
- Human-Computer Interaction
- Digital Health
Background:
- Eating behavior significantly impacts individual well-being, encompassing not just timing but also contextual factors like companions, location, and food type.
- Existing automated eating detection systems often neglect these crucial contextual elements.
Purpose of the Study:
- To develop a smartwatch-based system for detecting meal episodes using dominant hand movements.
- To design ecological momentary assessment (EMA) questions for capturing meal contexts.
- To validate the system's ability to trigger EMAs upon passive meal detection.
Main Methods:
- A smartwatch-based eating detection system was developed and deployed with 28 college students for 3 weeks.
- Ecological momentary assessment (EMA) questions were designed based on a survey of 162 students.
- The system passively detected meal episodes, triggering EMAs to collect contextual data.
Main Results:
- The novel meal detection system achieved high accuracy, detecting 96.48% of all meals consumed.
- Detection rates for specific meals were: 89.8% for breakfast, 99.0% for lunch, and 98.0% for dinner.
- Over 99% of detected meals were consumed with distractions, and 54.01% were eaten alone, highlighting potential links to unhealthy eating behaviors.
Conclusions:
- This is the first system to integrate EMA for capturing eating context, offering significant implications for well-being research.
- The gathered contextual data provides insights for designing personalized interventions to promote healthier eating habits.
