A Preclinical Human–Machine Reliability and Operational Safety Assessment Framework for a Rehabilitation Massage Robot Prototype

With the increasing use of massage robots as direct-contact rehabilitation devices, their evaluation must address both equipment failures and use-related risks. This study presents a prototype-stage assessment framework for a seated knee-adjustment bionic massage robot that integrates failure mode, effects, and criticality analysis (FMECA), task analysis, and fuzzy analytic hierarchy process (FAHP) weighting based on a fuzzy complementary judgment matrix. FMECA identified 28 failure modes across the actuation, sensing, and drive-control subsystems. Simulated-use testing involving 15 operators and 15 healthy adults acting as simulated patients comprised 19 observable subtasks and recorded 66 use errors across 285 subtask opportunities. The normalized scores for the actuation, sensing, drive-control, and operational-safety indicators were 0.796, 0.691, 0.808, and 0.768, respectively. Interval-based aggregation yielded Q = 0.848 (“Very Good”), whereas direct weighting of the same normalized inputs yielded Scont = 0.767, within the “Good” interval. Across η ∈ [0, 1], Q ranged from 0.753 to 0.902. A same-panel expert assessment yielded 0.904 and the same grade, although this comparison did not constitute independent validation. Inaccurate pressure-sensor output had the highest risk priority number of 32, while confusion between left- and right-side pressure readings was the most frequent use error, with 11 occurrences. The framework supports prototype-stage risk identification and prioritization but does not establish clinical efficacy, certified safety, or long-term reliability.