The Importance of Eating Speed Measurement

Eating speed has long been recognized as an important indicator in nutritional studies. Researchers have found that individuals who eat quickly are more likely to experience intake-related problems such as obesity, diabetes, and oral health issues. However, existing studies on eating speed have primarily relied on self-reported questionnaires, which are highly subjective and lack quantitative data.

A Novel Approach: Using Inertial Measurement Unit Sensors

In this groundbreaking study, a novel approach is proposed to measure eating speed in free-living environments automatically and objectively. The researchers utilize wrist-worn inertial measurement unit (IMU) sensors to gather data. These IMU sensors can detect specific gestures related to eating and drinking, allowing for the identification of individual bites.

Temporal Convolutional Network (TCN) and Multi-Head Attention Module (MHA)

To accurately identify bites from the IMU data, the researchers have developed a temporal convolutional network combined with a multi-head attention module (TCN-MHA). This powerful combination of algorithms ensures precise detection of eating gestures, enabling the determination of eating episodes.

Calculating Eating Speed

Once the bite sequences have been identified and clustered into eating episodes, the researchers calculate eating speed by dividing the time taken to finish the episode by the number of bites. This provides an objective and quantitative measure of an individual’s eating speed.

Validation and Results

The proposed approach is thoroughly validated using a 7-fold cross validation on the self-collected fine-annotated full-day-I (FD-I) dataset. Additionally, a hold-out experiment is conducted on the full-day-II (FD-II) dataset. These datasets, which are publicly available, consist of data collected from 61 participants in free-living environments, totaling 513 hours of observation.

The experimental results demonstrate the effectiveness of the proposed approach, achieving a mean absolute percentage error (MAPE) of 0.110 in the FD-I dataset and 0.146 in the FD-II dataset. These low error rates highlight the feasibility of automated eating speed measurement using IMU sensors.

Implications and Potential Future Research

This study is groundbreaking as it is the first to investigate automated eating speed measurement. By providing an objective and quantitative method of measuring eating speed, researchers can gain deeper insights into its relationship with various intake-related problems.

In the future, it would be interesting to explore the potential applications of this automated eating speed measurement approach. For instance, it could be integrated into wearable devices or mobile applications to provide individuals with real-time feedback on their eating speed. This could help individuals regulate their eating habits and improve their overall health.

In conclusion, this study presents a significant advancement in the measurement of eating speed. By utilizing IMU sensors and sophisticated algorithms, researchers have developed an automated and objective method for measuring eating speed in free-living environments. The findings of this study open up new possibilities for further research and potential interventions to address intake-related problems.

Read the original article