Behavioral and human-computer interaction requirements for a candidate abdominal-sound-enabled dietary self-regulation system: a two-phase study in Taiwanese university students.

BACKGROUND: Wearable nutrition tools frequently prioritize calorie surveillance and retrospective logging. However, university students may require low-friction support that aligns with everyday eating contexts. The hypothesis that abdominal sounds could serve as biologically plausible future gastrointestinal timing cues is well-founded. The purpose of this study was to define the behavioral and human-computer interaction (HCI) requirements for a candidate abdominal-sound-enabled dietary self-regulation system among Taiwanese university students. METHODS: The manuscript integrates two empirically linked phases. Phase 1 was a formative cross-sectional study of Taiwanese university students (n = 280; 45% male; mean age 22 ± 4 years) examining the Positive Eating Scale and its associations with eating behavior, diet quality, and BMI. Phase 2 was a randomized, formative 12-week intervention with 1:1 allocation (n = 200; n = 100 per group). A dynamic Kano analysis was employed to describe the evolution of HCI requirements over 12 weeks. RESULTS: The Positive Eating Scale exhibited a two-factor structure (KMO = 0.85; Bartlett's chi-square (28) = 14,605.46, p < 0.001), demonstrating satisfactory reliability for satisfaction (alpha = 0.85), pleasure (alpha = 0.92), and the total score (alpha = 0.87). The results indicated that satisfaction, rather than pleasure, exhibited a more favorable behavioral profile, characterized by lower restrained eating (r = -0.38), higher intuitive eating (r = 0.49), better perceived health (r = 0.30), and lower BMI (r = -0.25). At the group level, outcomes favored the intervention group for fruit-and-vegetable intake, unhealthy snack frequency, device adherence, and user satisfaction; these findings are descriptive and should not be interpreted as adjusted estimates of efficacy. The standardized mean differences calculated from the reported means and standard deviations were substantial. Initially, attractive features migrated to one-dimensional attributes, whereas several one-dimensional features became must-be requirements. CONCLUSIONS: The extant evidence supports the implementation of a non-punitive, satisfaction-centered, adaptive, and low-friction interface architecture for future wearable dietary self-regulation. Abdominal sounds should be interpreted solely as a candidate timing/context input until participant-level acoustic validation and downstream gut-brain-axis endpoints are obtained.