Age-friendly interaction design remains challenging in intelligent furniture applications, where complex interfaces and unreliable control may hinder middle-to-older adults’ safe operation and long-term acceptance. This study proposes a data-driven framework to support the design and evaluation of an age-friendly smart bed application by linking large-scale requirement mining, requirement-to-design translation, and controlled usability validation. First, large-scale e-commerce reviews (n = 20,174) were analyzed using LDA to identify dominant unmet needs. Key pain points clustered around interaction usability, information interpretability, and voice control reliability. Based on these findings, we developed an age-friendly smart bed app prototype. A within-subject usability study with 27 middle-to-older adults (55–65 years) compared the proposed system against a representative commercial reference application under standardized task protocols. The results showed significantly improved perceived usability, with a higher System Usability Scale (SUS) score (94.91 vs. 79.63, p < 0.001) and higher recommendation intention measured by overall NPS (59.52% vs. 34.39%). Beyond system-level usability improvement, this study makes three key contributions: (1) a reproducible ‘demand mining → design translation → empirical validation’ framework that bridges the methodological gaps in current intelligent furniture research; (2) operationalized mapping from LDA-derived user pain points to concrete multimodal interaction modules; and (3) empirical evidence from comparative usability testing with explicitly profiled middle-to-older adults (55–65 years) against a commercial reference system.
Song et al. (2026) studied this question.