Answers are grounded in this paper and cite their sources.
Current methods for assessing eating behaviour are often criticised for being unrepresentative of real-life eating experiences, involving resource-demanding subjective behavioural coding or reliance on self-report measures. The OCOsense glasses present a promising alternative by directly monitoring facial muscle movements without these limitations. Preliminary data suggested these glasses could detect several eating and chewing behaviours. However, this needed empirical validation: the purpose of this study. Forty-seven adults aged 18–33 (31 females and 16 males) participated in a 60-min lab-based breakfast, which was video recorded. Oral processing behaviours were annotated using manual behavioural coding on ELAN (version 6.7) and by analysis of sensor data from OCOsense glasses. Here, we focus on chewing data from intake of two breakfast foods: bagel and apple. For both foods, the number of chews recorded within each eating segment did not differ significantly between manual coded and glasses' algorithm output, and regression analyses revealed a strong correspondence between the two measures ( r (550) = 0.955). Likewise, mean estimated chewing rates did not differ significantly between the two methods. The participants' self-evaluation of their eating rate aligned well with recorded chewing behaviour and was equally detected by both methods. Overall, the OCOsense glasses correctly detected 81 % of eating and 84 % of non-eating behaviour. While there is scope for further refinement of the glasses' algorithm, overall facial muscle movements detected by OCOsense glasses were able to detect chewing movements for the two foods tested, and future work will expand this to other foods and eating behaviours outside of the laboratory.
No takes yet. Share an insight, caveat, or question.
Armitage et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: