FitNibble: A Field Study to Evaluate the Utility and Usability of Automatic Diet Monitoring in Food Journaling using an Eyeglasses-based Wearable

Title of the Paper

FitNibble: A Field Study to Evaluate the Utility and Usability of Automatic Diet Monitoring in Food Journaling Using an Eyeglasses-based Wearable

Paper Information

• Research Area: Wearable Devices and Automatic Diet Monitoring Systems
• Keywords: Diet monitoring, eating detection, food journaling, wearable devices, utility, usability, compliance, automatic diet monitoring

Research Background and Problems

• Identified Issues or Challenges:
  • The importance of diet monitoring for health is widely recognized, but existing methods often rely on self-reporting, which is difficult for average users to sustain.
  • Automatic Diet Monitoring systems (ADM) perform poorly in real-world scenarios outside the lab due to factors such as low ecological validity, poor social acceptability, and privacy concerns.
  • Diet monitoring tasks are complex (tracking "what to eat," "how much to eat," and "when to eat"), with current technologies primarily focusing on detecting "when to eat," while automation of food type and intake quantity remains a technical bottleneck.
• Significance of the Research:
  • There is an urgent need to develop a solution that makes food journaling as simple as using a smartwatch for step tracking, thereby enhancing user engagement and sustainability.
  • Improving the ecological validity, user utility, and social acceptability of diet monitoring systems will significantly advance their application in real-life scenarios, particularly in disease prevention and health improvement.
• Research Motivation and Related Work:
  • Previous studies have developed various ADM prototype systems, but lack evaluations of their utility and usability during long-term use in real-world settings.
  • This study is inspired by the recent eyeglasses-based FitByte system (capable of detecting eating and drinking behaviors), whose promising detection performance in free-living environments makes it a reference design for practical applications.

Solution

• Proposed Method or Solution:
  • This paper introduces an eyeglasses-based wearable ADM system—FitNibble, which can monitor users' eating behaviors in real time and push instant reminders at the start of a meal to assist in food journaling.
  • By improving the design of FitByte, the system incorporates sensors, a low-power Bluetooth module, and a smartphone application to form an end-to-end ADM system.
• Innovative Features of the Solution:
  • Real-time eating detection based on the latest machine learning models, demonstrating strong detection capabilities in free-living environments.
  • Compact and lightweight design that aligns with social acceptability and is suitable for daily wear.
  • Enhanced reminder functionality helps users remember to record dietary behaviors, overcoming the forgetfulness issue typical in manual journaling methods.
• Implementation Steps and Key Technologies:
  • Hardware: Install sensors (e.g., proximity sensors and inertial measurement units) on eyeglasses.
  • Software: Develop the FitNibble smartphone application to process data, trigger reminders, and interact with users.
  • Backend: Utilize deep neural network models to detect eating behaviors in real time from sensor data.

Research Findings

• Specific Results:
  • Through an 18-day field study with 13 participants, FitNibble significantly improved food journaling compliance rates (missed events reduced by 19.6%, p = 0.0132).
  • Using ADM reduced the perceived difficulty of journaling significantly (p = 0.005), and increased awareness of dietary behaviors, particularly snack consumption.
  • Test results showed that nearly half of the recorded events were prompted by system reminders, reflecting users' high reliance on FitNibble's notification functionality.
• Advantages Over Existing Solutions:
  • Compared to traditional manual journaling methods (or time-based reminders), FitNibble's instant notifications significantly reduced users' forgetfulness and cognitive burden.
  • The system also demonstrated good social acceptability and privacy awareness, with many users expressing willingness to adopt the device in daily life.
• Experiment or Evaluation Results:
  • Overall user compliance improved significantly: from a baseline of 54.7% to 77.4%.
  • Despite some users experiencing higher false positive rates (i.e., incorrect reminder triggers), the system's overall utility and potential benefits were still recognized.
  • User feedback surveys indicated positive reactions to the device's overall usage, with some users even observing behavioral changes in their dietary habits.
• Limitations and Future Directions:
  • The current study's sample size is relatively small, and the experimental design did not include a control group with balanced intervention order, potentially introducing sequence effects.
  • The hardware remains a prototype design; improving device miniaturization, waterproofing, and productization is crucial.
  • Future research should validate the findings in larger-scale randomized controlled trials, further optimize machine learning models to reduce false positive rates, and explore the feasibility of integrating camera functionality with privacy protection measures.

Additional Section

User demands for complete product forms, such as adding voice control and personalized reminder adjustments, as well as the acceptance of network camera functionality, were also raised and should be considered in future designs.