Wearable Monitoring of Ingestive Behavior

Wearable monitoring of ingestive behaviour encompasses the development of on-body devices and embedded sensors that capture the physiological and mechanical signals associated with eating and drinking. Technologies ranging from inertial sensors mounted on the wrist or jaw to optical and strain sensors in eyeglass frames enable real-time detection of chewing, swallowing and bite events without reliance on self-report. Advanced signal processing and machine-learning algorithms translate raw sensor outputs into granular metrics such as chew count, chewing rate and meal duration. These capabilities facilitate objective assessment of dietary intake, support personalised interventions for weight management and diabetes care, and inform epidemiological research into nutritional behaviour. By reducing participant burden and recall bias, wearable systems promise to enhance accuracy and ecological validity in both clinical and free-living settings, offering a scalable approach to monitoring global dietary patterns and informing public health strategies.

Research from Nature Portfolio

Recent studies have advanced non-invasive facial-signal sensing for monitoring ingestive behaviour. A notable approach embeds load cells in eyeglass hinges to amplify temporalis muscle oscillations during mastication. Pattern-recognition algorithms extract statistical features and classify chewing, talking and natural head movements with F1 scores exceeding 90% and overall accuracy above 89%, demonstrating the feasibility of unobtrusive, eyewear-based tracking of eating episodes.

Wearable Monitoring of Ingestive Behavior publication trend

The graph below shows the total number of articles in wearable monitoring of ingestive behavior across all publications each year (not limited to Nature Index journals).

Technical terms

Load cell: A transducer that converts mechanical force into a measurable electrical signal for detecting muscle-induced hinge movements.

Convolutional long short-term memory network: A deep-learning architecture combining convolutional layers for feature extraction with LSTM layers for temporal sequence modelling.

Hidden Markov model: A statistical model representing systems that transition between unobserved states, used to capture temporal dependencies in sensor outputs.

Accelerometer: An inertial sensor that measures linear acceleration and orientation changes, commonly used to detect hand-to-mouth gestures.

F1-score: A performance metric that balances precision and recall, indicating the accuracy of classification algorithms in detecting eating events.

References

1. Controlled and Real-Life Investigation of Optical Tracking Sensors in Smart Glasses for Monitoring Eating Behavior Using Deep Learning: Cross-Sectional Study. JMIR mHealth and uHealth (2024).
2. Automatic, wearable-based, in-field eating detection approaches for public health research: a scoping review. npj Digital Medicine (2020).
3. Automatic Measurement of Chew Count and Chewing Rate during Food Intake. Electronics (2016).
4. A glasses-type wearable device for monitoring the patterns of food intake and facial activity. Scientific Reports (2017).