Reliability and Discriminant Ability of an Instrumented Timed Up and Go Test in People With Postsurgical Orthopedic Conditions: Quantitative Study

Background: The Timed Up and Go (TUG) test is widely used to assess mobility and fall risk in older adults and orthopedic patients. Its instrumented variant (iTUG), based on inertial measurement units, enables an objective quantification of motor performance and can even be implemented using smartphone technology. However, its broader clinical adoption remains limited by concerns about reliability, feasibility, and the interpretability of the extracted parameters. Objective: This study aimed to evaluate the test-retest reliability of variables derived from a single-sensor iTUG in orthopedic inpatients undergoing rehabilitation and to determine whether a subset of reliable sensor-based metrics can support a multidimensional assessment of functional mobility and discriminate among common orthopedic conditions. Methods: We recruited 104 inpatients at discharge from a rehabilitation ward after total hip arthroplasty, total knee arthroplasty, or femur fracture. Each participant performed the iTUG test on 2 consecutive days using a smartphone-based solution consisting of an inertial measurement unit placed on the lower back. From 100 extracted variables, those with excellent test-retest reliability (intraclass correlation coefficient >= 0.75) were retained. Exploratory factor analysis was used to identify underlying mobility domains, and linear discriminant analysis with 10-fold cross-validation tested their ability to classify diagnostic groups. Results: Out of 100 iTUG-derived variables, 36 demonstrated excellent test-retest reliability, and 25 were retained for multivariate analysis. Exploratory factor analysis identified 5 factors-walking ability, pace or rhythm, sit-to-walk smoothness, turning ability, and mediolateral angular stability-explaining 80.8% of the total variance. These factors showed good classification accuracy (68%) and achieved an area under the curve of 0.86 and an overall mean accuracy of 0.68 (SD 0.14) for distinguishing among total hip arthroplasty, total knee arthroplasty, and femur fracture. In contrast, total iTUG duration alone yielded an area under the curve of 0.62. All patients used walking aids, and gait variables were more reliable than jerk-based or coordination metrics. Conclusions: The single-sensor iTUG provides reliable and clinically informative metrics that go beyond traditional stopwatch timing, enabling a multidimensional view of functional mobility in orthopedic patients. The approach is feasible, interpretable, and compatible with real-world mobile health apps, supporting personalized rehabilitation monitoring and future integration into digital decision support systems.