Evaluating the Validity and Feasibility of a Smartwatch-based Eating Detection System to Passively and Automatically Detect Eating Events in Child-parent Dyads

The study will: 1) determine whether the smartwatch-based system accurately detects eating events in child-parent dyads in controlled settings and 2) evaluate the feasibility and practicality of passively detecting eating events in child-parent dyads over 3 days in free-living settings. The study will include two phases. During the laboratory visit, child-parent dyads will wear the smartwatch on their dominant hand and perform activities including eating gestures. These activities will be recorded with a video camera, and the videos will be coded for the ground truth times of eating. In the second phase of the study, child-parent dyads will continue wearing the smartwatch for 3 more days in free-living conditions. In the free-living period, parents will receive personalized Ecological Momentary Assessment (EMA) prompts reminding them to activate the smartwatch.

• Smartwatch and EMA-based eating behavior tracking (Other): Participants (child-parent dyads) will wear a smartwatch on their dominant hand during a laboratory session and for three days in free-living conditions. In the lab, dyads will perform eating-related activities (e.g., eating with utensils, eating with hands, drinking) and non-eating activities (e.g., walking, writing, brushing teeth) while being video recorded for ground truth validation. Parents will receive a 20-minute training on using Ecological Momentary Assessment (EMA) prompts to record meal and snack times and will respond to EMA reminders during the free-living period. Adherence will be monitored through smartwatch wear time and EMA response rates.

Primary outcomes

• Accuracy and Performance of Smartwatch-based Eating Detection System in Controlled Conditions [During the controlled laboratory session (Day 1).]: Accuracy and performance of smartwatch-based system for detecting eating and drinking events compared to video-coded ground truth during a controlled laboratory conditions. Metrics include: Precision (proportion of correctly detected eating events among all detected events), Recall/Sensitivity (proportion of actual eating events detected), Specificity (proportion of non-eating events correctly identified as non-eating) F1-score (harmonic mean of precision and recall), Overall accuracy (proportion of correct classification).
• Adherence to Smartwatch Protocol [During the free-living phase (Days 2-4).]: Proportion of families meeting adherence criteria, defined as both parent and child wearing the smartwatch during eating on at least 2 of 3 days and the parent responding to ≥75% of EMA prompts related to smartwatch charging and app functionality.

Sex: All, ages: 8 Years to 12 Years, accepts healthy volunteers

Inclusion Criteria:

* Parents or caregivers (18-70 years) who have children aged 8-12 years
* Child is willing and able to wear smartwatch during school hours (have not restrictions in the school setting)

Exclusion Criteria:

* Any condition or circumstance that could impede study completion
* Child does not follow a regular eating pattern
* Child eats less than 1 meal and 1 snack in a day
* Child is restricted or allergic to the study foods
* Refusal or unable to use the smartwatch to collect data for the 3-day period in free - living conditions
* Parental refusal or unable to respond Ecological Momentary Assessment prompts

• Hanim E Diktas, PhD, Principal investigator, Pennington Biomedical Research Center