Wearable abdominal acoustic biofeedback for dietary self-regulation among Taiwanese university students: an exploratory randomized pilot study.

Background: Manual dietary self-monitoring can impose a burden on university students, limiting sustained engagement with digital nutrition tools. AbSound was developed as a wearable abdominal-acoustic biofeedback system that converts candidate gastrointestinal sound events into uncertainty-aware reflective prompts, rather than validated diagnoses of hunger, satiety, dietary intake, diet quality, or eating speed. This exploratory, randomized pilot study assessed feasibility, acceptability, user experience, preliminary self-reported behavioral signals, and agreement in acoustic event detection among Taiwanese university students. Method: Sixty first- and second-year university students in Taipei were randomized 1:1 to AbSound plus a companion app or to an active app-based dietary-tracking comparator for 4 weeks. Primary feasibility outcomes included retention, daily wear time, days worn, prompt exposure, prompt acknowledgment, reported burden, and adverse events. Exploratory behavioral outcomes included self-reported late-night eating, perceived eating speed, a brief dietary-quality score, self-regulation of eating behavior, weight, and waist circumference. TAM ratings, Kano feature classifications, brief interviews, and descriptive acceptance-path analyses characterized adoption and design priorities. Acoustic performance was evaluated Results: All participants completed follow-up. AbSound users wore the device for 9.2 ± 2.2 h per day on 25.7 ± 2.0 of 28 study days, corresponding to approximately 91.8% of possible study days. Prompt acknowledgment averaged 78% ± 9%. Selected, manually reviewed segments showed approximately 88% agreement in event detection, but this estimate does not establish physiological-state validation. Compared with the active comparator, AbSound users showed larger exploratory reductions in self-reported late-night eating. They reported lower perceived eating-speed scores, higher brief dietary-quality scores, and a marginally greater increase in eating self-regulation. Weight and waist circumference showed no clinically interpretable between-group differences. Acceptance ratings were high; comfort and privacy emerged as must-be design attributes, while feedback agreement and timeliness were performance attributes. Conclusion: AbSound demonstrated short-term feasibility and acceptability as a reflective abdominal-acoustic biofeedback tool for non-clinical Taiwanese university students. The findings support further user-centered development but do not establish clinical efficacy or validated detection of nutritional or physiological states. Larger preregistered trials require independent acoustic validation, objective or repeated dietary reference measures, longer follow-up, and explicit psychological-safety monitoring.