TL;DR
This paper introduces the SPLENDID eating detection sensor, an ear-worn device using an air microphone and PPG sensor to objectively measure eating events. While technically promising for improving food intake monitoring, user feedback highlights that comfort and discreetness are critical for real-world adoption and require further development.
What Problem Does This Paper Address?
Accurate monitoring of food intake is crucial for various health applications, including weight management and obesity prevention. However, current methods, primarily relying on self-reporting (e.g., food diaries, recall), are prone to bias and incomplete data due to underreporting, forgetfulness, and the need for conscious effort. While technological advancements have led to devices that can automatically detect eating events, there's a lack of practical, user-friendly solutions that are acceptable for everyday use. The SPLENDID project aims to bridge this gap by developing an ear-worn device that can objectively capture eating events, thereby reducing reliance on self-reporting and providing more accurate data.
Methods
The development of the SPLENDID eating detection sensor followed an iterative, incremental approach, with user feedback integrated at each stage. The core technology involves an ear-worn sensor unit combining:
- Air Microphone: Detects chewing sounds.
- Photoplethysmogram (PPG) Sensor: Measures blood volume changes in ear tissue, which are affected by chewing.
Initially, preliminary prototypes were tested in a lab setting. Subsequent stages involved developing a wearable version with a separate "datalogger" (housing a data acquisition system, battery, and accelerometer) connected via a cable. The final integrated version incorporated the sensor, datalogger, and a smartphone app for user interaction and feedback, along with a webtool for data overview and goal setting.
User evaluation was conducted in four stages:
- Concept Evaluation: Interviews with health professionals and a focus group with potential end-users to gauge initial perceptions.
- Preliminary Prototype Evaluation: Laboratory testing with young adults to assess initial comfort and functionality.
- First Wearable Version Evaluation: A semi-controlled study with overweight adults over two days.
- Integrated Version Evaluation: A 4-week feasibility study with overweight adults using the full system.
Datasets
Not discussed in the paper. The studies involved participants recruited for specific evaluation stages, rather than using pre-existing datasets.
Key Insights and Ideas
- User Enthusiasm vs. Practicality: While potential users are generally enthusiastic about the concept of an objective eating detection sensor, their willingness to use it hinges heavily on comfort and discreetness. This highlights that technical accuracy alone is insufficient for widespread adoption.
- The "Forgettable" Device: A recurring theme is the need for users to "forget" they are wearing the device for it to be truly effective in capturing natural eating patterns. This implies that any discomfort or noticeability will lead to conscious awareness, defeating the purpose of objective, unobtrusive monitoring.
- Ear-worn Advantage: The ear-worn form factor is seen as advantageous because it can be mistaken for listening devices, offering a degree of discreetness compared to other wearable sensor placements.
- Iterative Development is Key: The multi-stage evaluation process, incorporating feedback from both health professionals and end-users, was crucial in identifying and addressing user-centric challenges throughout the development lifecycle.
Key Results
- Wearer Comfort is a Major Hurdle: In the final 4-week feasibility study, the eating detection sensor received an average wearer comfort grade of 3.7 out of 10. Discomfort was the primary reason participants wore the sensor for less than 2 hours a day.
- Limited Usage Due to Discomfort: Participants in the final study used the sensor for an average of 19 out of 28 instructed days, and when used, it was for an average of 1.9 hours per day. This is significantly less than the intended continuous wear.
- Technical Performance Promising (but not the focus here): The paper references previous work indicating good technical performance (e.g., accuracy of 0.938 in earlier versions), but this study focuses on user perception.
- Mixed Social Reactions: Participants reported mixed reactions from their social environment regarding the sensor's noticeability, scoring "People in my environment noticed the sensor" and "I did not like it when people noticed the sensor" at an average of 3.4 out of 7.
Limitations and Future Work
- Wearer Comfort and Discreetness: The most significant limitation identified is the lack of sufficient wearer comfort and discreetness. Participants experienced discomfort after an average of 1 hour and 20 minutes, leading to reduced usage.
- Limited Usage Duration: The study's feasibility was impacted by the need to limit wear time due to discomfort, which may have prevented participants from fully forgetting they were wearing the device, thus affecting the naturalness of their eating behavior.
- Future Work: The authors suggest further development should focus on improving the hardware design to enhance wearer comfort and discreetness. Options include offering different shapes and sizes, or making the device more akin to hearing aids. Repeating the feasibility study with improved versions is recommended. Additionally, addressing technical issues and clarifying the added value for users who don't feel the need for such a reminder are areas for future consideration.
Summary updated